PG_AUTO_FAILOVER(1)
✨ NAME ✨
pg_auto_failover — and yeah bestie, this whole page is the pg_auto_failover Documentation. 💅
so here's the tea: pg_auto_failover is an Open Source Software project. the development happens at https://github.com/citusdata/pg_auto_failover and it is public: literally everyone is welcome to participate by opening issues or pull requests, giving feedback, etc. 🫶
and remember, the first steps are to actually play with the pg_autoctl command, then read the entire available documentation (after all, I took the time to write it 😭), and then to address the community in a kind and polite way — the same way you would expect people to use when addressing you. periodt.
NOTE: The development of pg_auto_failover has been driven by Citus Data, since then a team at Microsoft. The Citus Data team at Microsoft generously maintains the pg_auto_failover Open Source Software so that its users may continue using it in production. For enhancements, improvements, and new features, consider contributing to the project. Pull Requests are reviewed as part of the offered maintenance.
NOTE: Assistance is provided as usual with Open Source projects, on a voluntary basis. If you need help to cook a patch, enhance the documentation, or even to use the software, you're welcome to ask questions and expect some level of free guidance.
🌸 INTRODUCTION TO PG_AUTO_FAILOVER 🌸
pg_auto_failover is an extension for PostgreSQL that monitors and manages failover for postgres clusters. it is optimised for simplicity and correctness. no cap, that's the whole vibe.
🦋 Single Standby Architecture 🦋
[image: pg_auto_failover Architecture with a primary and a standby node] [image] pg_auto_failover architecture with a primary and a standby node.
pg_auto_failover implements Business Continuity for your PostgreSQL services. it implements a single PostgreSQL service using multiple nodes with automated failover, and automates PostgreSQL maintenance operations in a way that guarantees availability of the service to its users and applications. 💖
to that end, pg_auto_failover uses three nodes (machines, servers) per PostgreSQL service:
- a PostgreSQL primary node,
- a PostgreSQL secondary node, using Synchronous Hot Standby,
- a
pg_auto_failoverMonitor node that acts both as a witness and an orchestrator.
the pg_auto_failover Monitor implements a state machine and relies on in-core PostgreSQL facilities to deliver HA. for example: when the secondary node is detected to be unavailable, or when its lag is reported above a defined threshold (the default is 1 WAL files, or 16MB, see the pgautofailover.promote_wal_log_threshold GUC on the pg_auto_failover monitor), then the Monitor removes it from the synchronous_standby_names setting on the primary node. until the secondary is back to being monitored healthy, failover and switchover operations are not allowed, preventing data loss. 🔥
💜 Multiple Standby Architecture 💜
[image: pg_auto_failover Architecture for a standalone PostgreSQL service] [image] pg_auto_failover architecture with a primary and two standby nodes.
in the pictured architecture, pg_auto_failover implements Business Continuity and data availability by implementing a single PostgreSQL service using multiple with automated failover and data redundancy. even after losing any Postgres node in a production system, this architecture maintains two copies of the data on two different nodes. 🌈
when using more than one standby, different architectures can be achieved with pg_auto_failover, depending on the objectives and trade-offs needed for your production setup.
⭐ Multiple Standbys Architecture with 3 standby nodes, one async ⭐
[image: pg_auto_failover architecture with a primary and three standby nodes] [image] pg_auto_failover architecture with a primary and three standby nodes.
when setting the three parameters above, it's possible to design very different Postgres architectures for your production needs.
in this case, the system is setup with two standby nodes participating in the replication quorum, allowing for number_sync_standbys = 1. the system always maintains a minimum of two copies of the data set: one on the primary, another one on either node B or node C. whenever we lose one of those nodes, we can hold to this guarantee of two copies of the data set. ✨
adding to that, we have the standby server D which has been set up to not participate in the replication quorum. node D will not be found in the synchronous_standby_names list of nodes. also, node D is set up in a way to never be a candidate for failover, with candidate-priority = 0.
this architecture would fit a situation where nodes A, B, and C are deployed in the same data center or availability zone, and node D in another. those three nodes are set up to support the main production traffic and implement high availability of both the Postgres service and the data set.
node D might be set up for Business Continuity in case the first data center is lost, or maybe for reporting the need for deployment on another application domain. 🦋
🎀 Citus Architecture 🎀
[image: pg_auto_failover architecture with a Citus formation] [image] pg_auto_failover architecture with a Citus formation.
pg_auto_failover implements Business Continuity for your Citus services. pg_auto_failover implements a single Citus formation service using multiple Citus nodes with automated failover, and automates PostgreSQL maintenance operations in a way that guarantees availability of the service to its users and applications.
in that case, pg_auto_failover knows how to orchestrate a Citus coordinator failover and a Citus worker failover. a Citus worker failover can be achieved with a very minimal downtime to the application, where during a short time window SQL writes may error out. 😭
in this figure we see a single standby node for each Citus node, coordinator and workers. it is possible to implement more standby nodes, and even read-only nodes for load balancing, see Citus Secondaries and read-replica. 👑
✨ QUICK START ✨
okay so bestie, pg_auto_failover ships with this command line tool called pg_autoctl and it literally has SO many commands for managing your Postgres nodes. like, a whole wardrobe. and to build the Postgres architectures described in this documentation (and honestly more), you can usually get away with just some of those many pg_autoctl commands. not all 40 of them, bestie, breathe.
this section is a lil short intro to the main commands that are useful when you're getting started with pg_auto_failover. there are more commands available that help deal with a variety of situations — see the Manual Pages for the whole list, because we're not gatekeeping.
to understand which replication settings to use in your case, see the Architecture Basics section and then the Multi-node Architectures section.
to follow a step by step guide that you can reproduce on your own Azure subscription and create a production Postgres setup from VMs, see the Azure VMs Tutorial section. to get started with a local setup using docker compose to run multiple Postgres nodes, see the Tutorial section.
to understand how to setup pg_auto_failover in a way that is compliant with your internal security guide lines, read the Security settings for pg_auto_failover section. slay.
💻 Command line environment, configuration files, etc 💻
as a command line tool pg_autoctl depends on some environment variables. mostly, the tool re-uses the Postgres environment variables that you might already know. it's giving "reuse, recycle, iconic".
to manage a Postgres node pg_auto_failover needs to know its data directory location on-disk. for that, some users will find it easier to export the PGDATA variable in their environment. the alternative consists of always using the --pgdata option that is available to all the pg_autoctl commands. pick your fighter. 🥊
🌱 Creating Postgres Nodes 🌱
to get started with the simplest Postgres failover setup, 3 nodes are needed: the pg_auto_failover monitor, and 2 Postgres nodes that will get assigned roles by the monitor. one Postgres node will be assigned the primary role, the other one will get assigned the secondary role. the way that this is literally a whole little trio. 🫶
to create the monitor use the command:
$ pg_autoctl create monitorthe create the Postgres nodes use the following command on each node you want to create:
$ pg_autoctl create postgreswhile those create commands initialize your nodes, now you have to actually run the Postgres service that are expected to be running. for that you can manually run the following command on every node:
$ pg_autoctl runit is also possible (and recommended) to integrate the pg_auto_failover service in your usual service management facility. when using systemd the following commands can be used to produce the unit file configuration required:
$ pg_autoctl show systemd
INFO HINT: to complete a systemd integration, run the following commands:
INFO pg_autoctl -q show systemd --pgdata "/tmp/pgaf/m" | sudo tee /etc/systemd/system/pgautofailover.service
INFO sudo systemctl daemon-reload
INFO sudo systemctl enable pgautofailover
INFO sudo systemctl start pgautofailover
[Unit]
...while it is expected that for a production deployment each node actually is a separate machine (virtual or physical, or even a container), it is also possible to run several Postgres nodes all on the same machine for testing or development purposes. lowkey convenient.
TIP: when running severalpg_autoctlnodes on the same machine for testing or contributing topg_auto_failover, each Postgres instance needs to run on its own port, and with its own data directory. it can make things easier to then set the environment variablesPGDATAandPGPORTin each terminal, shell, or tab where each instance is started. no cap. 🧚
🔍 Inspecting nodes 🔍
once your Postgres nodes have been created, and once each pg_autoctl service is running, it is possible to inspect the current state of the formation with the following command:
$ pg_autoctl show statethe pg_autoctl show state commands outputs the current state of the system only once. sometimes it would be nice to have an auto-updated display such as provided by common tools such as watch(1) or top(1) and the like. for that, the following commands are available (see also pg_autoctl watch):
$ pg_autoctl watch
$ pg_autoctl show state --watchto analyze what's been happening to get to the current state, it is possible to review the past events generated by the pg_auto_failover monitor with the following command:
$ pg_autoctl show eventsHINT: thepg_autoctl showcommands can be run from any node in your system. those command need to connect to the monitor and print the current state or the current known list of events as per the monitor view of the system. usepg_autoctl show state --localto have a view of the local state of a given node without connecting to the monitor Postgres instance. the option--jsonis available in mostpg_autoctlcommands and switches the output format from a human readable table form to a program friendly JSON pretty-printed output. bestie that's the good stuff. ✨
🎛️ Inspecting and Editing Replication Settings 🎛️
when creating a node it is possible to use the --candidate-priority and the --replication-quorum options to set the replication properties as required by your choice of Postgres architecture.
to review the current replication settings of a formation, use one of the two following commands, which are convenient aliases (the same command with two ways to invoke it):
$ pg_autoctl show settings
$ pg_autoctl get formation settingsit is also possible to edit those replication settings at any time while your nodes are in production: you can change your mind or adjust to new elements without having to re-deploy everything. just use the following commands to adjust the replication settings on the fly:
$ pg_autoctl set formation number-sync-standbys
$ pg_autoctl set node replication-quorum
$ pg_autoctl set node candidate-priorityIMPORTANT: thepg_autoctl getandpg_autoctl setcommands always connect to the monitor Postgres instance. thepg_autoctl setcommand then changes the replication settings on the node registration on the monitor. then the monitor assigns theAPPLY_SETTINGSstate to the current primary node in the system for it to apply the new replication settings to its Postgres streaming replication setup. as a result, thepg_autoctl setcommands requires a stable state in the system to be allowed to proceed. namely, the current primary node in the system must have both its Current State and its Assigned State set toprimary, as per thepg_autoctl show stateoutput. periodt. 💅
🛠️ Implementing Maintenance Operations 🛠️
when a Postgres node must be taken offline for a maintenance operation, such as e.g. a kernel security upgrade or a minor Postgres update, it is best to make it so that the pg_auto_failover monitor knows about it. communicate, bestie. 🗣️
- for one thing, a node that is known to be in maintenance does not participate in failovers. if you are running with two Postgres nodes, then failover operations are entirely prevented while the standby node is in maintenance.
- moreover, depending on your replication settings, enabling maintenance on your standby ensures that the primary node switches to async replication before Postgres is shut down on the secondary, avoiding write queries to be blocked.
to implement maintenance operations, use the following commands:
$ pg_autoctl enable maintenance
$ pg_autoctl disable maintenancethe main pg_autoctl run service that is expected to be running in the background should continue to run during the whole maintenance operation. when a node is in the maintenance state, the pg_autoctl service is not controlling the Postgres service anymore.
note that it is possible to enable maintenance on a primary Postgres node, and that operation then requires a failover to happen first. it is possible to have pg_auto_failover orchestrate that for you when using the command:
$ pg_autoctl enable maintenance --allow-failoverIMPORTANT: thepg_autoctl enableandpg_autoctl disablecommands requires a stable state in the system to be allowed to proceed. namely, the current primary node in the system must have both its Current State and its Assigned State set toprimary, as per thepg_autoctl show stateoutput. no cap, don't skip this. 😤
👑 Manual failover, switchover, and promotions 👑
in the cases when a failover is needed without having an actual node failure, the pg_auto_failover monitor can be used to orchestrate the operation. use one of the following commands, which are synonyms in the pg_auto_failover design:
$ pg_autoctl perform failover
$ pg_autoctl perform switchoverfinally, it is also possible to "elect" a new primary node in your formation with the command:
$ pg_autoctl perform promotionIMPORTANT: thepg_autoctl performcommands requires a stable state in the system to be allowed to proceed. namely, the current primary node in the system must have both its Current State and its Assigned State set toprimary, as per thepg_autoctl show stateoutput. main character energy only. 🌟
🌈 What's next? 🌈
this section of the documentation is meant to help users get started by focusing on the main commands of the pg_autoctl tool. each command has many options that can have very small impact, or pretty big impact in terms of security or architecture. read the rest of the manual to understand how to best use the many pg_autoctl options to implement your specific Postgres production architecture. you got this bestie, fr fr. 💖
🦋 TUTORIAL 🦋
in this guide we'll create a Postgres setup with two nodes, a primary and a standby. then we'll add a second standby node. we'll simulate failure in the Postgres nodes and see how the system continues to function. it's giving resilience. 💅
this tutorial uses docker compose in order to separate the architecture design from some of the implementation details. this allows reasoning at the architecture level within this tutorial, and better see which software component needs to be deployed and run on which node.
the setup provided in this tutorial is good for replaying at home in the lab. it is not intended to be production ready though. in particular, no attention have been spent on volume management. after all, this is a tutorial: the goal is to walk through the first steps of using pg_auto_failover to implement Postgres automated failover. we're just vibing, not shipping. 🎀
📋 Pre-requisites 📋
when using docker compose we describe a list of services, each service may run on one or more nodes, and each service just runs a single isolated process in a container.
within the context of a tutorial, or even a development environment, this matches very well to provisioning separate physical machines on-prem, or Virtual Machines either on-prem on in a Cloud service. iconic. ✨
So the docker image we're using in this tutorial is called pg_auto_failover:tutorial, and you can build it locally with the Dockerfile that ships inside the GitHub repo for pg_auto_failover 💖
Want to build it? Two ways, bestie: either use the Makefile that's provided and run make build, or just run the docker build command directly:
$ git clone https://github.com/citusdata/pg_auto_failover
$ cd pg_auto_failover/docs/tutorial
$ docker compose build✨ Postgres failover with two nodes ✨
docker compose makes it genuinely easy to spin up an architecture that looks like this:
[image: pg_auto_failover architecture with a primary and a standby node] 🦋
This kind of setup gives you failover capabilities, but lowkey it does not give you High Availability of both the Postgres service and the data. Wanna understand more about that? Go read the Multi-node Architectures chapter of the docs, bestie, it's all in there.
To create a cluster we use this docker compose definition:
x-node: &node
build:
context: ../..
dockerfile: Dockerfile
tags:
- pg_auto_failover:tutorial
volumes:
- /var/lib/postgres
environment:
PGDATA: /var/lib/postgres/pgaf
PGUSER: tutorial
PGDATABASE: tutorial
PG_AUTOCTL_HBA_LAN: true
PG_AUTOCTL_AUTH_METHOD: "trust"
PG_AUTOCTL_SSL_SELF_SIGNED: true
PG_AUTOCTL_MONITOR: "postgresql://autoctl_node@monitor/pg_auto_failover"
expose:
- 5432
services:
app:
build:
context: .
dockerfile: Dockerfile.app
environment:
PGUSER: tutorial
PGDATABASE: tutorial
PGHOST: node1,node2
PGPORT: 5432
PGAPPNAME: tutorial
PGSSLMODE: require
PGTARGETSESSIONATTRS: read-write
monitor:
image: pg_auto_failover:tutorial
volumes:
- /var/lib/postgres
environment:
PGDATA: /var/lib/postgres/pgaf
PG_AUTOCTL_SSL_SELF_SIGNED: true
expose:
- 5432
command: |
pg_autoctl create monitor --auth trust --run
node1:
<<: *node
hostname: node1
command: |
pg_autoctl create postgres --name node1 --run
node2:
<<: *node
hostname: node2
command: |
pg_autoctl create postgres --name node2 --run
node3:
<<: *node
hostname: node3
command: |
pg_autoctl create postgres --name node3 --runTo run the whole Citus cluster with HA from this definition, you just hit:
$ docker compose up app monitor node1 node2That command above starts the services up. First service is the monitor, and it's created with the command pg_autoctl create monitor. The options for this command are exposed in the environment — but ngl, they could've been given on the command line too:
$ pg_autoctl create postgres --ssl-self-signed --auth trust --pg-hba-lan --runWhile the Postgres nodes are getting provisioned by docker compose, you can run the command below and get yourself a dynamic dashboard to follow everything happening in real time. It's basically top, but for pg_auto_failover 👑
$ docker compose exec monitor pg_autoctl watchAfter a little while you can run the pg_autoctl show state command and see a stable result:
$ docker compose exec monitor pg_autoctl show state
Name | Node | Host:Port | TLI: LSN | Connection | Reported State | Assigned State
------+-------+------------+----------------+--------------+---------------------+--------------------
node2 | 1 | node2:5432 | 1: 0/3000148 | read-write | primary | primary
node1 | 2 | node1:5432 | 1: 0/3000148 | read-only | secondary | secondaryAnd we can review the available Postgres URIs with the pg_autoctl show uri command 💅
$ docker compose exec monitor pg_autoctl show uri
Type | Name | Connection String
-------------+---------+-------------------------------
monitor | monitor | postgres://autoctl_node@58053a02af03:5432/pg_auto_failover?sslmode=require
formation | default | postgres://node2:5432,node1:5432/tutorial?target_session_attrs=read-write&sslmode=require🌸 Add application data 🌸
Okay so now let's create a database schema with a single table, and put some data in there.
$ docker compose exec app psql
-- in psql
CREATE TABLE companies
(
id bigserial PRIMARY KEY,
name text NOT NULL,
image_url text,
created_at timestamp without time zone NOT NULL,
updated_at timestamp without time zone NOT NULL
);Next, download and ingest some sample data — still from inside our psql session:
\copy companies from program 'curl -o- https://examples.citusdata.com/mt_ref_arch/companies.csv' with csv
( COPY 75 )🔥 Our first failover 🔥
Okay bestie, here's the slay part: with pg_auto_failover it's possible (and easy!) to trigger a failover without having to orchestrate an incident or power down the current primary. No chaos required 💖
$ docker compose exec monitor pg_autoctl perform switchover
14:57:16 992 INFO Waiting 60 secs for a notification with state "primary" in formation "default" and group 0
14:57:16 992 INFO Listening monitor notifications about state changes in formation "default" and group 0
14:57:16 992 INFO Following table displays times when notifications are received
Time | Name | Node | Host:Port | Current State | Assigned State
---------+-------+-------+------------+---------------------+--------------------
14:57:16 | node2 | 1 | node2:5432 | primary | draining
14:57:16 | node1 | 2 | node1:5432 | secondary | prepare_promotion
14:57:16 | node1 | 2 | node1:5432 | prepare_promotion | prepare_promotion
14:57:16 | node1 | 2 | node1:5432 | prepare_promotion | stop_replication
14:57:16 | node2 | 1 | node2:5432 | primary | demote_timeout
14:57:17 | node2 | 1 | node2:5432 | draining | demote_timeout
14:57:17 | node2 | 1 | node2:5432 | demote_timeout | demote_timeout
14:57:19 | node1 | 2 | node1:5432 | stop_replication | stop_replication
14:57:19 | node1 | 2 | node1:5432 | stop_replication | wait_primary
14:57:19 | node2 | 1 | node2:5432 | demote_timeout | demoted
14:57:19 | node2 | 1 | node2:5432 | demoted | demoted
14:57:19 | node1 | 2 | node1:5432 | wait_primary | wait_primary
14:57:19 | node2 | 1 | node2:5432 | demoted | catchingup
14:57:26 | node2 | 1 | node2:5432 | demoted | catchingup
14:57:38 | node2 | 1 | node2:5432 | demoted | catchingup
14:57:39 | node2 | 1 | node2:5432 | catchingup | catchingup
14:57:39 | node2 | 1 | node2:5432 | catchingup | secondary
14:57:39 | node2 | 1 | node2:5432 | secondary | secondary
14:57:40 | node1 | 2 | node1:5432 | wait_primary | primary
14:57:40 | node1 | 2 | node1:5432 | primary | primaryThe new state after the failover looks like this:
$ docker compose exec monitor pg_autoctl show state
Name | Node | Host:Port | TLI: LSN | Connection | Reported State | Assigned State
------+-------+------------+----------------+--------------+---------------------+--------------------
node2 | 1 | node2:5432 | 2: 0/5002698 | read-only | secondary | secondary
node1 | 2 | node1:5432 | 2: 0/5002698 | read-write | primary | primaryAnd we can verify that we still have the data available 🫶
docker compose exec app psql -c "select count(*) from companies"
count
-------
75
(1 row)🌈 Multiple Standbys Architectures 🌈
The docker-compose.yml file comes with a third node that you can bring up to get the following architecture:
[image: pg_auto_failover architecture with a primary and two standby nodes]
🎀 Adding a second standby node 🎀
To run a second standby node — or a third Postgres node, honestly, same energy — just run:
$ docker compose up -d node3We can see the resulting replication settings with this command:
$ docker compose exec monitor pg_autoctl show settings
Context | Name | Setting | Value
----------+---------+---------------------------+-------------------------------------------------------------
formation | default | number_sync_standbys | 1
primary | node1 | synchronous_standby_names | 'ANY 1 (pgautofailover_standby_1, pgautofailover_standby_3)'
node | node2 | candidate priority | 50
node | node1 | candidate priority | 50
node | node3 | candidate priority | 50
node | node2 | replication quorum | true
node | node1 | replication quorum | true
node | node3 | replication quorum | true🧚 Editing the replication settings while in production 🧚
It's then possible to change the production architecture you got by playing around with the Replication Settings and Postgres Architectures commands. Specifically, try this command to change the candidate_priority of node3 to zero, so it never becomes a candidate for failover:
$ docker compose exec node3 pg_autoctl set candidate-priority 0 --name node3To see the replication settings for all the nodes, this command is super useful — and it's described in more detail in the Replication Settings and Postgres Architectures section:
$ docker compose exec monitor pg_autoctl show settings Context | Name | Setting | Value
----------+---------+---------------------------+-------------------------------------------------------------
formation | default | number_sync_standbys | 1
primary | node1 | synchronous_standby_names | 'ANY 1 (pgautofailover_standby_1, pgautofailover_standby_3)'
node | node2 | candidate priority | 50
node | node1 | candidate priority | 50
node | node3 | candidate priority | 0
node | node2 | replication quorum | true
node | node1 | replication quorum | true
node | node3 | replication quorum | trueokay bestie so then, in a separate terminal — but in the same directory, because that's just how docker compose does projects — you can run this little watch command:
$ docker compose exec monitor pg_autoctl watchand back in your main terminal, while the watch output is right there being visible, you can trigger a switchover:
$ docker compose exec monitor pg_autoctl perform switchovergetting familiar with those commands is literally one of your next steps 💅 the manual has full coverage of them at these links:
pg_autoctl watchpg_autoctl show statepg_autoctl show settingspg_autoctl set node candidate-priority
🧹 Cleanup 🧹
to yeet the entire tutorial environment, you just run:
$ docker compose downgone. poof. no notes. 💀
🦋 Next steps 🦋
like we said at the top of this tutorial, the way we're using docker compose here is not meant to be production ready, okay? it's just genuinely useful for understanding and playing with a distributed system like a multi-node Postgres setup with failovers. that's the whole vibe. 🎀
see the command pg_autoctl do tmux compose session for more details on how to run a docker compose test environment with docker compose — including external volumes for each node. very much iykyk.
also see the complete Azure VMs Tutorial for a guide on how-to provision an Azure network and then Azure VMs with pg_auto_failover, including systemd coverage and a failover triggered by stopping a whole entire VM. which is… literally the next section, so let's go 🫶
🌈 AZURE VMS TUTORIAL 🌈
in this guide we're gonna create a primary and a secondary Postgres node and set up pg_auto_failover to replicate data between them. then we simulate failure in the primary node and watch the system smoothly switch (fail over) to the secondary. iconic behavior honestly ✨
for illustration we'll run our databases on virtual machines in the Azure platform, but the techniques here are relevant to any cloud provider or on-premise network, no cap. we'll use four virtual machines: a primary database, a secondary database, a monitor, and an "application." the monitor watches the other nodes' health, manages global state, and assigns nodes their roles. main character energy for real 👑
☁️ Create virtual network ☁️
our database machines need to talk to each other and to the monitor node, so let's create a virtual network:
az group create \
--name ha-demo \
--location eastus
az network vnet create \
--resource-group ha-demo \
--name ha-demo-net \
--address-prefix 10.0.0.0/16we need to open ports 5432 (Postgres) and 22 (SSH) between the machines, and also give ourselves access from our remote IP. we do this with a network security group and a subnet:
az network nsg create \
--resource-group ha-demo \
--name ha-demo-nsg
az network nsg rule create \
--resource-group ha-demo \
--nsg-name ha-demo-nsg \
--name ha-demo-ssh-and-pg \
--access allow \
--protocol Tcp \
--direction Inbound \
--priority 100 \
--source-address-prefixes `curl ifconfig.me` 10.0.1.0/24 \
--source-port-range "*" \
--destination-address-prefix "*" \
--destination-port-ranges 22 5432
az network vnet subnet create \
--resource-group ha-demo \
--vnet-name ha-demo-net \
--name ha-demo-subnet \
--address-prefixes 10.0.1.0/24 \
--network-security-group ha-demo-nsgfinally, add four virtual machines (ha-demo-a, ha-demo-b, ha-demo-monitor, and ha-demo-app). for speed we background the az vm create processes and run them in parallel:
# create VMs in parallel
for node in monitor a b app
do
az vm create \
--resource-group ha-demo \
--name ha-demo-${node} \
--vnet-name ha-demo-net \
--subnet ha-demo-subnet \
--nsg ha-demo-nsg \
--public-ip-address ha-demo-${node}-ip \
--image debian \
--admin-username ha-admin \
--generate-ssh-keys &
done
waitto make it easier to SSH into these VMs in future steps, let's make a shell function that retrieves their IP addresses:
# run this in your local shell as well
vm_ip () {
az vm list-ip-addresses -g ha-demo -n ha-demo-$1 -o tsv \
--query '[] [] .virtualMachine.network.publicIpAddresses[0].ipAddress'
}
# for convenience with ssh
for node in monitor a b app
do
ssh-keyscan -H `vm_ip $node` >> ~/.ssh/known_hosts
donelet's review what we created so far, bestie:
az resource list --output table --query \
"[?resourceGroup=='ha-demo'].{ name: name, flavor: kind, resourceType: type, region: location }"and this shows the following resources:
Name ResourceType Region
------------------------------- ----------------------------------------------------- --------
ha-demo-a Microsoft.Compute/virtualMachines eastus
ha-demo-app Microsoft.Compute/virtualMachines eastus
ha-demo-b Microsoft.Compute/virtualMachines eastus
ha-demo-monitor Microsoft.Compute/virtualMachines eastus
ha-demo-appVMNic Microsoft.Network/networkInterfaces eastus
ha-demo-aVMNic Microsoft.Network/networkInterfaces eastus
ha-demo-bVMNic Microsoft.Network/networkInterfaces eastus
ha-demo-monitorVMNic Microsoft.Network/networkInterfaces eastus
ha-demo-nsg Microsoft.Network/networkSecurityGroups eastus
ha-demo-a-ip Microsoft.Network/publicIPAddresses eastus
ha-demo-app-ip Microsoft.Network/publicIPAddresses eastus
ha-demo-b-ip Microsoft.Network/publicIPAddresses eastus
ha-demo-monitor-ip Microsoft.Network/publicIPAddresses eastus
ha-demo-net Microsoft.Network/virtualNetworks eastuslook at all that. slay. 💖
📦 Install the "pg_autoctl" executable 📦
this guide uses Debian Linux, but similar steps work on other distributions too — all that differs are the packages and the paths. see Installing pg_auto_failover.
the pg_auto_failover system is distributed as a single pg_autoctl binary with subcommands to initialize and manage a replicated PostgreSQL service. we'll install the binary with the operating system package manager on all nodes. it'll help us run and observe PostgreSQL:
for node in monitor a b app
do
az vm run-command invoke \
--resource-group ha-demo \
--name ha-demo-${node} \
--command-id RunShellScript \
--scripts \
"sudo touch /home/ha-admin/.hushlogin" \
"curl https://install.citusdata.com/community/deb.sh | sudo bash" \
"sudo DEBIAN_FRONTEND=noninteractive apt-get install -q -y postgresql-common" \
"echo 'create_main_cluster = false' | sudo tee -a /etc/postgresql-common/createcluster.conf" \
"sudo DEBIAN_FRONTEND=noninteractive apt-get install -q -y postgresql-17-auto-failover" \
"sudo usermod -a -G postgres ha-admin" &
done
wait👀 Run a monitor 👀
the pg_auto_failover monitor is the first component to run. it periodically attempts to contact the other nodes and watches their health. it also maintains global state that "keepers" on each node consult to determine their own roles in the system:
# on the monitor virtual machine
ssh -l ha-admin `vm_ip monitor` -- \
pg_autoctl create monitor \
--auth trust \
--ssl-self-signed \
--pgdata monitor \
--pgctl /usr/lib/postgresql/17/bin/pg_ctlthis command initializes a PostgreSQL cluster at the location pointed to by the --pgdata option. when --pgdata is omitted, pg_autoctl attempts to use the PGDATA environment variable. and if a PostgreSQL instance already existed in the destination directory, this command would have configured it to serve as a monitor instead.
pg_auto_failover installs the pgautofailover Postgres extension, and grants access to a new autoctl_node user.
in the Quick Start we use --auth trust to dodge complex security settings. heads up though: the Postgres trust authentication method is not considered a reasonable choice for production environments. consider either using the --skip-pg-hba option or --auth scram-sha-256 and then setting up passwords yourself. don't be out here trusting randos in prod 🙈
at this point the monitor is created. now we install it as a service with systemd so it'll resume if the VM restarts:
ssh -T -l ha-admin `vm_ip monitor` << CMD
pg_autoctl -q show systemd --pgdata ~ha-admin/monitor > pgautofailover.service
sudo mv pgautofailover.service /etc/systemd/system
sudo systemctl daemon-reload
sudo systemctl enable pgautofailover
sudo systemctl start pgautofailover
CMD🚀 Bring up the nodes 🚀
we'll create the primary database using the pg_autoctl create subcommand:
ssh -l ha-admin `vm_ip a` -- \
pg_autoctl create postgres \
--pgdata ha \
--auth trust \
--ssl-self-signed \
--username ha-admin \
--dbname appdb \
--hostname ha-demo-a.internal.cloudapp.net \
--pgctl /usr/lib/postgresql/17/bin/pg_ctl \
--monitor 'postgres://autoctl_node@ha-demo-monitor.internal.cloudapp.net/pg_auto_failover?sslmode=require'notice the user and database name in the monitor connection string — those are literally what monitor init created. we also give it the path to pg_ctl so that the keeper will use the correct version of pg_ctl in the future, even if other versions of postgres end up installed on the system. future-proofing, we love to see it ✨
In the example above, the keeper creates a primary database. It's giving main character energy — it picks node A as the primary because the monitor reports there are literally no other nodes in the system yet. And this is one example of how the keeper is state-based: it makes observations, then adjusts its state, in this case going from "init" to "single." 🦋
Also bestie, we need to add a setting to trust connections from our "application" VM:
ssh -T -l ha-admin `vm_ip a` << CMD
echo 'hostssl "appdb" "ha-admin" ha-demo-app.internal.cloudapp.net trust' \
>> ~ha-admin/ha/pg_hba.conf
CMDAt this point the monitor and primary node are created and running. ✨ Next we need to run the keeper. It's an independent process so that it can continue operating even if the PostgreSQL process goes terminates on the node. We'll install it as a service with systemd so that it will resume if the VM restarts. 💅
ssh -T -l ha-admin `vm_ip a` << CMD
pg_autoctl -q show systemd --pgdata ~ha-admin/ha > pgautofailover.service
sudo mv pgautofailover.service /etc/systemd/system
sudo systemctl daemon-reload
sudo systemctl enable pgautofailover
sudo systemctl start pgautofailover
CMDNext, connect to node B and do the same process. We'll do both steps at once:
ssh -l ha-admin `vm_ip b` -- \
pg_autoctl create postgres \
--pgdata ha \
--auth trust \
--ssl-self-signed \
--username ha-admin \
--dbname appdb \
--hostname ha-demo-b.internal.cloudapp.net \
--pgctl /usr/lib/postgresql/17/bin/pg_ctl \
--monitor 'postgres://autoctl_node@ha-demo-monitor.internal.cloudapp.net/pg_auto_failover?sslmode=require'
ssh -T -l ha-admin `vm_ip b` << CMD
pg_autoctl -q show systemd --pgdata ~ha-admin/ha > pgautofailover.service
sudo mv pgautofailover.service /etc/systemd/system
sudo systemctl daemon-reload
sudo systemctl enable pgautofailover
sudo systemctl start pgautofailover
CMDIt discovers from the monitor that a primary exists, and then switches its own state to be a hot standby and begins streaming WAL contents from the primary. 🔥
🌈 Node communication 🌈
For convenience, pg_autoctl modifies each node's pg_hba.conf file to allow the nodes to connect to one another. For instance, pg_autoctl added the following lines to node A:
# automatically added to node A
hostssl "appdb" "ha-admin" ha-demo-a.internal.cloudapp.net trust
hostssl replication "pg_auto_failover_replicator" ha-demo-b.internal.cloudapp.net trust
hostssl "appdb" "pg_auto_failover_replicator" ha-demo-b.internal.cloudapp.net trustFor pg_hba.conf on the monitor node, pg_autoctl inspects the local network and makes its best guess about the subnet to allow. In our case it guessed correctly:
# automatically added to the monitor
hostssl "pg_auto_failover" "autoctl_node" 10.0.1.0/24 trustHeads up bestie: if worker nodes have more ad-hoc addresses and are not in the same subnet, it's better to disablepg_autoctl's automatic modification of pg_hba using the--skip-pg-hbacommand line option during creation. You will then need to edit the hba file by hand. Another reason for manual edits would be to use special authentication methods. 🫶
👀 Watch the replication 👀
First let's verify that the monitor knows about our nodes, and see what states it has assigned them:
ssh -l ha-admin `vm_ip monitor` pg_autoctl show state --pgdata monitor
Name | Node | Host:Port | LSN | Reachable | Current State | Assigned State
-------+-------+--------------------------------------+-----------+-----------+---------------------+--------------------
node_1 | 1 | ha-demo-a.internal.cloudapp.net:5432 | 0/3000060 | yes | primary | primary
node_2 | 2 | ha-demo-b.internal.cloudapp.net:5432 | 0/3000060 | yes | secondary | secondaryThis looks good. We can add data to the primary, and later see it appear in the secondary. We'll connect to the database from inside our "app" virtual machine, using a connection string obtained from the monitor. ✨
ssh -l ha-admin `vm_ip monitor` pg_autoctl show uri --pgdata monitor
Type | Name | Connection String
-----------+---------+-------------------------------
monitor | monitor | postgres://autoctl_node@ha-demo-monitor.internal.cloudapp.net:5432/pg_auto_failover?sslmode=require
formation | default | postgres://ha-demo-b.internal.cloudapp.net:5432,ha-demo-a.internal.cloudapp.net:5432/appdb?target_session_attrs=read-write&sslmode=requireNow we'll get the connection string and store it in a local environment variable:
APP_DB_URI=$( \
ssh -l ha-admin `vm_ip monitor` \
pg_autoctl show uri --formation default --pgdata monitor \
)The connection string contains both our nodes, comma separated, and includes the url parameter ?target_session_attrs=read-write telling psql that we want to connect to whichever of these servers supports reads and writes. That will be the primary server. 👑
# connect to database via psql on the app vm and
# create a table with a million rows
ssh -l ha-admin -t `vm_ip app` -- \
psql "'$APP_DB_URI'" \
-c "'CREATE TABLE foo AS SELECT generate_series(1,1000000) bar;'"💀 Cause a failover 💀
Now that we've added data to node A, let's switch which is considered the primary and which the secondary. After the switch we'll connect again and query the data, this time from node B.
# initiate failover to node B
ssh -l ha-admin -t `vm_ip monitor` \
pg_autoctl perform switchover --pgdata monitorOnce node B is marked "primary" (or "wait_primary") we can connect and verify that the data is still present:
# connect to database via psql on the app vm
ssh -l ha-admin -t `vm_ip app` -- \
psql "'$APP_DB_URI'" \
-c "'SELECT count(*) FROM foo;'"It shows
count
---------
1000000😭 Cause a node failure 😭
This plot is too boring, time to introduce a problem. We'll turn off VM for node B (currently the primary after our previous failover) and watch node A get promoted. 🎀
In one terminal let's keep an eye on events:
ssh -t -l ha-admin `vm_ip monitor` -- \
watch -n 1 -d pg_autoctl show state --pgdata monitorIn another terminal we'll turn off the virtual server.
az vm stop \
--resource-group ha-demo \
--name ha-demo-bAfter a number of failed attempts to talk to node B, the monitor determines the node is unhealthy and puts it into the "demoted" state. The monitor promotes node A to be the new primary.
Name | Node | Host:Port | LSN | Reachable | Current State | Assigned State
-------+-------+--------------------------------------+-----------+-----------+---------------------+--------------------
node_1 | 1 | ha-demo-a.internal.cloudapp.net:5432 | 0/6D4E068 | yes | wait_primary | wait_primary
node_2 | 2 | ha-demo-b.internal.cloudapp.net:5432 | 0/6D4E000 | yes | demoted | catchingupNode A cannot be considered in full "primary" state since there is no secondary present, but it can still serve client requests. It is marked as "wait_primary" until a secondary appears, to indicate that it's running without a backup. 🙈
Let's add some data while B is offline.
# notice how $APP_DB_URI continues to work no matter which node
# is serving as primary
ssh -l ha-admin -t `vm_ip app` -- \
psql "'$APP_DB_URI'" \
-c "'INSERT INTO foo SELECT generate_series(1000001, 2000000);'"🧚 Resurrect node B 🧚
Run this command to bring node B back online:
az vm start \
--resource-group ha-demo \
--name ha-demo-bNow the next time the keeper retries its health check, it brings the node back. Node B goes through the state "catchingup" while it updates its data to match A. Once that's done, B becomes a secondary, and A is now a full primary again. slay ✨
Name | Node | Host:Port | LSN | Reachable | Current State | Assigned State
-------+-------+--------------------------------------+------------+-----------+---------------------+--------------------
node_1 | 1 | ha-demo-a.internal.cloudapp.net:5432 | 0/12000738 | yes | primary | primary
node_2 | 2 | ha-demo-b.internal.cloudapp.net:5432 | 0/12000738 | yes | secondary | secondaryWhat's more, if we connect directly to the database again, all two million rows are still present. 🔥
ssh -l ha-admin -t `vm_ip app` -- \
psql "'$APP_DB_URI'" \
-c "'SELECT count(*) FROM foo;'"It shows
count
---------
2000000🛠️ INSTALLING PG_AUTO_FAILOVER 🛠️
okay bestie, we got you native system packages for pg_auto_failover on most of the popular Linux distros out there. we're not gonna make you suffer through a build-from-source situation unless you truly want to 💅
Use the steps below to install pg_auto_failover on PostgreSQL 17. At the current time pg_auto_failover is compatible with PostgreSQL 13 to 17.
🐧 Ubuntu or Debian
📦 Postgres apt repository
Binary packages for debian and its derivatives (that's ubuntu, bestie) are available from the apt.postgresql.org repository — install by following the linked documentation and then:
$ sudo apt-get install pg-auto-failover-cli
$ sudo apt-get install postgresql-17-auto-failoverThe Postgres extension named "pgautofailover" is only necessary on the monitor node. To install that extension, you can install the postgresql-17-auto-failover package when using Postgres 17. It's available for other Postgres versions too. ✨
🙈 Avoiding the default Postgres service
When installing the debian Postgres package, the installation script will initialize a Postgres data directory automatically, and register it to the systemd services. When using pg_auto_failover, it is best to avoid that step. (like, it's not wrong exactly, but it's giving chaos energy and you don't want that)
To avoid automated creation of a Postgres data directory when installing the debian package, follow those steps:
$ curl https://www.postgresql.org/media/keys/ACCC4CF8.asc | apt-key add -
$ echo "deb http://apt.postgresql.org/pub/repos/apt bookworm-pgdg main" > /etc/apt/sources.list.d/pgdg.list
# bypass initdb of a "main" cluster
$ echo 'create_main_cluster = false' | sudo tee -a /etc/postgresql-common/createcluster.conf
$ apt-get update
$ apt-get install -y --no-install-recommends postgresql-17That way when it's time to pg_autoctl create monitor or pg_autoctl create postgres there is no confusion about how to handle the default Postgres service created by debian: it has not been created at all. clean. iconic. no notes. 🎀
🎩 Fedora, CentOS, or Red Hat
The Postgres community packaging team for RPM based system has worked on supporting pg_auto_failover. Binary packages are available by following the documentation at PostgreSQL Yum Repository.
A single package named pg_auto_failover is available on the RPM based systems, containing both the monitor Postgres extension and the pg_autoctl command line. one package, both things, we love a bundle 🫶
⚙️ Installing a pgautofailover Systemd unit
The command pg_autoctl show systemd outputs a systemd unit file that you can use to setup a boot-time registered service for pg_auto_failover on your machine.
Here's a sample output from the command:
$ export PGDATA=/var/lib/postgresql/monitor
$ pg_autoctl show systemd
13:44:34 INFO HINT: to complete a systemd integration, run the following commands:
13:44:34 INFO pg_autoctl -q show systemd --pgdata "/var/lib/postgresql/monitor" | sudo tee /etc/systemd/system/pgautofailover.service
13:44:34 INFO sudo systemctl daemon-reload
13:44:34 INFO sudo systemctl start pgautofailover
[Unit]
Description = pg_auto_failover
[Service]
WorkingDirectory = /var/lib/postgresql
Environment = 'PGDATA=/var/lib/postgresql/monitor'
User = postgres
ExecStart = /usr/lib/postgresql/17/bin/pg_autoctl run
Restart = always
StartLimitBurst = 0
[Install]
WantedBy = multi-user.targetCopy/pasting the commands given in the hint output from the command will enable the pgautofailer service on your system, when using systemd. yes bestie, literally just copy-paste, the tool did the homework for you 📋✨
It is important that PostgreSQL is started by pg_autoctl rather than by systemd itself, as it might be that a failover has been done during a reboot, for instance, and that once the reboot complete we want the local Postgres to re-join as a secondary node where it used to be a primary node. 🚨 so like, don't get cute and start it yourself, fr fr.
🏗️ Building pg_auto_failover from sources
To build the project, make sure you have installed the build-dependencies, then just type make. You can install the resulting binary using make install.
For this to work please consider adding both the binary and the source repositories to your debian distribution by using the following apt sources, as an example targetting the debian bullseye distribution:
deb http://apt.postgresql.org/pub/repos/apt bullseye-pgdg main
deb-src http://apt.postgresql.org/pub/repos/apt bullseye-pgdg mainThen we can install the build dependencies for Postgres, knowing that pg_auto_failover uses the same build dependencies:
$ sudo apt-get build-dep -y --no-install-recommends postgresql-17Then build pg_auto_failover from sources with the following instructions:
$ make -s clean && make -s -j12 all
$ sudo make -s install🧠 ARCHITECTURE BASICS 🧠
pg_auto_failover is designed as a simple and robust way to manage automated Postgres failover in production. On-top of robust operations, pg_auto_failover setup is flexible and allows either Business Continuity or High Availability configurations. pg_auto_failover design includes configuration changes in a live system without downtime. no downtime?? in this economy?? we love to see it 😭👑
pg_auto_failover is designed to be able to handle a single PostgreSQL service using three nodes. In this setting, the system is resilient to losing any one of three nodes.
[image: pg_auto_failover Architecture for a standalone PostgreSQL service] [image] pg_auto_failover Architecture for a standalone PostgreSQL service.
It is important to understand that when using only two Postgres nodes then pg_auto_failover is optimized for Business Continuity. In the event of losing a single node, pg_auto_failover is capable of continuing the PostgreSQL service, and prevents any data loss when doing so, thanks to PostgreSQL Synchronous Replication.
That said, there is a trade-off involved in this architecture. The business continuity bias relaxes replication guarantees for asynchronous replication in the event of a standby node failure. This allows the PostgreSQL service to accept writes when there's a single server available, and opens the service for potential data loss if the primary server were also to fail. ⚠️ okay so this is the part where you read carefully bestie — it's a real trade-off, not a vibe, it's a genuine "you might lose data" situation if the primary ALSO fails.
🕹️ The pg_auto_failover Monitor
Each PostgreSQL node in pg_auto_failover runs a Keeper process which informs a central Monitor node about notable local changes. Some changes require the Monitor to orchestrate a correction across the cluster:
- New nodes — At initialization time, it's necessary to prepare the configuration of each node for PostgreSQL streaming replication, and get the cluster to converge to the nominal state with both a primary and a secondary node in each group. The monitor determines each new node's role. ✨
- Node failure — The monitor orchestrates a failover when it detects an unhealthy node. The design of
pg_auto_failoverallows the monitor to shut down service to a previously designated primary node without causing a"split-brain"situation. (no split-brain, we simply do not allow it 💅)
The monitor is the authoritative node that manages global state and makes changes in the cluster by issuing commands to the nodes' keeper processes. A pg_auto_failover monitor node failure has limited impact on the system. While it prevents reacting to other nodes' failures, it does not affect replication. The PostgreSQL streaming replication setup installed by pg_auto_failover does not depend on having the monitor up and running. lowkey reassuring ngl 🦋
📖 pg_auto_failover Glossary
pg_auto_failover handles a single PostgreSQL service with the following concepts:
👀 Monitor
The pg_auto_failover monitor is a service that keeps track of one or several formations containing groups of nodes.
The monitor is implemented as a PostgreSQL extension, so when you run the command pg_autoctl create monitor a PostgreSQL instance is initialized, configured with the extension, and started. The monitor service embeds a PostgreSQL instance. it's a Postgres in a Postgres, inception-core 🌀
🗂️ Formation
A formation is a logical set of PostgreSQL services that are managed together.
It is possible to operate many formations with a single monitor instance. Each formation has a group of Postgres nodes and the FSM orchestration implemented by the monitor applies separately to each group.
👯 Group
A group of two PostgreSQL nodes work together to provide a single PostgreSQL service in a Highly Available fashion. A group consists of a PostgreSQL primary server and a secondary server setup with Hot Standby synchronous replication. Note that pg_auto_failover can orchestrate the whole setting-up of the replication for you. (you literally don't have to touch it, it does it, slay)
In pg_auto_failover versions up to 1.3, a single Postgres group can contain only two Postgres nodes. Starting with pg_auto_failover 1.4, there's no limit to the number of Postgres nodes in a single group. Note that each Postgres instance that belongs to the same group serves the same dataset in its data directory (PGDATA). 📁
NOTE: The notion of a formation that contains multiple groups in pg_auto_failover is useful when setting up and managing a whole Citus formation, where the coordinator nodes belong to group zero of the formation, and each Citus worker node becomes its own group and may have Postgres standby nodes.
🐕 Keeper
The pg_auto_failover keeper is an agent that must be running on the same server where your PostgreSQL nodes are running. The keeper controls the local PostgreSQL instance (using both the pg_ctl command-line tool and SQL queries), and communicates with the monitor:
- it sends updated data about the local node, such as the WAL delta in between servers, measured via PostgreSQL statistics views. 📊
- it receives state assignments from the monitor. 📥
Also the keeper maintains local state that includes the most recent communication established with the monitor and the other PostgreSQL node of its group, enabling it to detect Network Partitions. 🕵️
NOTE: Inpg_auto_failoverversions up to and including 1.3, the keeper process started withpg_autoctl runmanages a separate Postgres instance, running as its own process tree. Starting inpg_auto_failoverversion 1.4, the keeper process (started withpg_autoctl run) runs the Postgres instance as a sub-process of the mainpg_autoctlprocess, allowing tighter control over the Postgres execution. Running the sub-process also makes the solution work better both in container environments (because it's now a single process tree) and with systemd, because it uses a specific cgroup per service unit. 🐳✨
🖥️ Node
A node is a server (virtual or physical) that runs PostgreSQL instances and a keeper service. At any given time, any node might be a primary or a secondary Postgres instance. The whole point of pg_auto_failover is to decide this state.
As a result, refrain from naming your nodes with the role you intend for them. Their roles can change. If they didn't, your system wouldn't need pg_auto_failover! 💀
so like, don't name it primary-1, bestie. that's not it. main character energy only, not fixed-role energy. 👑
🌸 State 🌸
Okay bestie, a state is literally just the representation of the per-instance and per-group situation. That's it, that's the vibe. 💅
The monitor and the keeper implement a Finite State Machine to drive operations in the PostgreSQL groups, which is what lets pg_auto_failover implement High Availability with the goal of zero data loss. Iconic behavior honestly. ✨
The keeper main loop enforces the current expected state of the local PostgreSQL instance, and reports the current state and some more information to the monitor. The monitor uses this set of information and its own health-check information to drive the State Machine and assign a goal state to the keeper.
And then — plot twist — the keeper implements the transitions between a current state and a monitor-assigned goal state. It's giving glow-up arc, fr fr. 🦋
🎀 Client-side HA 🎀
Implementing client-side High Availability is included in PostgreSQL's driver libpq from version 10 onward. Using this driver, it is possible to specify multiple host names or IP addresses in the same connection string:
$ psql -d "postgresql://host1,host2/dbname?target_session_attrs=read-write"
$ psql -d "postgresql://host1:port2,host2:port2/dbname?target_session_attrs=read-write"
$ psql -d "host=host1,host2 port=port1,port2 target_session_attrs=read-write"When using either of the syntax above, the psql application attempts to connect to host1, and when successfully connected, checks the target_session_attrs as per the PostgreSQL documentation of it:
If this parameter is set toread-write, only a connection in which read-write transactions are accepted by default is considered acceptable. The querySHOW transaction_read_onlywill be sent upon any successful connection; if it returnson, the connection will be closed. If multiple hosts were specified in the connection string, any remaining servers will be tried just as if the connection attempt had failed. The default value of this parameter,any, regards all connections as acceptable.
So when the connection attempt to host1 fails, or when the target_session_attrs can not be verified, then the psql application attempts to connect to host2. She's persistent, we love that. 💖
The behavior is implemented in the connection library libpq, so any application using it can benefit from this implementation, not just psql. Not psql gatekeeping the good stuff, periodt. 👑
When using pg_auto_failover, configure your application connection string to use the primary and the secondary server host names, and set target_session_attrs=read-write too, so that your application automatically connects to the current primary, even after a failover occurred. No cap, this is the move. 🔥
🔍 Monitoring protocol 🔍
The monitor interacts with the data nodes in 2 ways:
- Data nodes periodically connect and run
SELECT pgautofailover.node_active(...)to communicate their current state and obtain their goal state. - The monitor periodically connects to all the data nodes to see if they are healthy, doing the equivalent of
pg_isready.
When a data node calls node_active, the state of the node is stored in the pgautofailover.node table and the state machines of both nodes are progressed. The state machines are described later in this readme. The monitor typically only moves one state forward and waits for the node(s) to converge except in failure states.
If a node is not communicating to the monitor, it will either cause a failover (if node is a primary), disabling synchronous replication (if node is a secondary), or cause the state machine to pause until the node comes back (other cases). In most cases, the latter is harmless, though in some cases it may cause downtime to last longer, e.g. if a standby goes down during a failover. 😭 Not the downtime staying longer than invited.
To simplify operations, a node is only considered unhealthy if the monitor cannot connect and it hasn't reported its state through node_active for a while. This allows, for example, PostgreSQL to be restarted without causing a health check failure. Thoughtful queen behavior honestly. 🫶
🧚 Synchronous vs. asynchronous replication 🧚
By default, pg_auto_failover uses synchronous replication, which means all writes block until at least one standby node has reported receiving them. To handle cases in which the standby fails, the primary switches between two states called wait_primary and primary based on the health of standby nodes, and based on the replication setting number_sync_standby.
When in the wait_primary state, synchronous replication is disabled by automatically setting synchronous_standby_names = '' to allow writes to proceed. However doing so also disables failover, since the standby might get arbitrarily far behind. If the standby is responding to health checks and within 1 WAL segment of the primary (by default), synchronous replication is enabled again on the primary by setting synchronous_standby_names = '*' which may cause a short latency spike since writes will then block until the standby has caught up. Not the latency spike showing up unannounced 😭 but it's fine, it's fine.
When using several standby nodes with replication quorum enabled, the actual setting for synchronous_standby_names is set to a list of those standby nodes that are set to participate to the replication quorum.
If you wish to disable synchronous replication, you need to add the following to postgresql.conf:
synchronous_commit = 'local'This ensures that writes return as soon as they are committed on the primary — under all circumstances. In that case, failover might lead to some data loss, but failover is not initiated if the secondary is more than 10 WAL segments (by default) behind on the primary. During a manual failover, the standby will continue accepting writes from the old primary. The standby will stop accepting writes only if it's fully caught up (most common), the primary fails, or it does not receive writes for 2 minutes.
💀 Heads up bestie: disabling synchronous replication means failover might lead to data loss. Read that part twice.
⭐ A note about performance ⭐
In some cases the performance impact on write latency when setting synchronous replication makes the application fail to deliver expected performance. If testing or production feedback shows this to be the case, it is beneficial to switch to using asynchronous replication.
The way to use asynchronous replication in pg_auto_failover is to change the synchronous_commit setting. This setting can be set per transaction, per session, or per user. It does not have to be set globally on your Postgres instance. Per-user? Per-session? Flexible. She contains multitudes. 💅
One way to benefit from that would be:
alter role fast_and_loose set synchronous_commit to local;That way performance-critical parts of the application don't have to wait for the standby nodes. Only use this when you can also lower your data durability guarantees.
🙈 The role name is genuinely fast_and_loose and honestly? Respect. But fr fr, only do this if you're cool with weaker durability.
🌈 Node recovery 🌈
When bringing a node back after a failover, the keeper (pg_autoctl run) can simply be restarted. It will also restart postgres if needed and obtain its goal state from the monitor. If the failed node was a primary and was demoted, it will learn this from the monitor. Once the node reports, it is allowed to come back as a standby by running pg_rewind. If it is too far behind, the node performs a new pg_basebackup. Welcome back bestie, we missed you 🫶
👑 MULTI-NODE ARCHITECTURES 👑
Pg_auto_failover allows you to have more than one standby node, and offers advanced control over your production architecture characteristics. More nodes, more control, more main character energy. ✨
🦋 Architectures with two standby nodes 🦋
When adding your second standby node with default settings, you get the following architecture:
[image: pg_auto_failover architecture with two standby nodes] [image]
pg_auto_failover architecture with two standby nodes.UNINDENTIn this case, three nodes get set up with the same characteristics, achieving HA for both the Postgres service and the production dataset. An important setting for this architecture is number_sync_standbys.
The replication setting number_sync_standbys sets how many standby nodes the primary should wait for when committing a transaction. In order to have a good availability in your system, pg_auto_failover requires number_sync_standbys + 1 standby nodes participating in the replication quorum: this allows any standby node to fail without impact on the system's ability to respect the replication quorum.
When only two nodes are registered in a group on the monitor we have a primary and a single secondary node. Then number_sync_standbys can only be set to zero. When adding a second standby node to a pg_auto_failover group, then the monitor automatically increments number_sync_standbys to one, as we see in the diagram above.
When number_sync_standbys is set to zero then pg_auto_failover implements the Business Continuity setup as seen in Architecture Basics: synchronous replication is then used as a way to guarantee that failover can be implemented without data loss.
In more details:
- With
number_sync_standbysset to one, this architecture always maintains two copies of the dataset: one on the current primary node (node A in the previous diagram), and one on the standby that acknowledges the transaction first (either node B or node C in the diagram).
When one of the standby nodes is unavailable, the second copy of the dataset can still be maintained thanks to the remaining standby.
When both the standby nodes are unavailable, then it's no longer possible to guarantee the replication quorum, and thus writes on the primary are blocked. The Postgres primary node waits until at least one standby node acknowledges the transactions locally committed, thus degrading your Postgres service to read-only. 😭 Not the read-only downgrade but okay, safety first.
- It is possible to manually set
number_sync_standbysto zero when having registered two standby nodes to the monitor, overriding the default behavior.
In that case, when the second standby node becomes unhealthy at the same time as the first standby node, the primary node is assigned the state Wait_primary. In that state, synchronous replication is disabled on the primary by setting synchronous_standby_names to an empty string. Writes are allowed on the primary, even though there's no extra copy of the production dataset available at this time.
Setting number_sync_standbys to zero allows data to be written even when both standby nodes are down. In this case, a single copy of the production data set is kept and, if the primary was then to fail, some data will be lost. How much depends on your backup and recovery mechanisms.
💀 Two standbys down and no quorum? Writes get blocked and your service goes read-only. That's the whole point of the quorum, bestie.
🌸 Replication Settings and Postgres Architectures 🌸
The entire flexibility of pg_auto_failover can be leveraged with the following three replication settings:
- Number of sync stanbys
- Replication quorum
- Candidate priority
💖 Number Sync Standbys 💖
This parameter is used by Postgres in the synchronous_standby_names parameter: number_sync_standby is the number of synchronous standbys for whose replies transactions must wait.
This parameter can be set at the formation level in pg_auto_failover, which lowkey means it applies to whoever is the current primary right now, and then it "follows" a failover so it applies to whatever new primary replaces them. Iconic, honestly. 💅
To set this parameter to the value <n>, you run:
pg_autoctl set formation number-sync-standbys <n>The default value in pg_auto_failover is zero. And when it's zero, the Postgres parameter synchronous_standby_names can be either '*' or '', like so:
synchronous_standby_names = '*'— bestie, this means any standby may participate in the replication quorum for transactions withsynchronous_commitset toonor higher values.
pg_autofailover uses synchronous_standby_names = '*' when there's at least one standby that is known to be healthy.
synchronous_standby_names = ''(empty string) disables synchronous commit and makes all your commits asynchronous, meaning transaction commits will not wait for replication. In other words: a single copy of your production data is maintained whensynchronous_standby_namesis set that way. 😭
pg_autofailover uses synchronous_standby_names = '' only when number_sync_standbys is set to zero and there's no standby node known healthy by the monitor.
Okay so for number_sync_standbys to be a non-zero value, pg_auto_failover requires that at least number_sync_standbys + 1 standby nodes be registered in the system. Math is math. ✨
When the first standby node gets added to the pg_auto_failover monitor, the only acceptable value for number_sync_standbys is zero. Then when a second standby is added that participates in the replication quorum, number_sync_standbys is automatically set to one. The system is doing it for you, fr fr.
The command pg_autoctl set formation number-sync-standbys can be used to change the value of this parameter in a formation, even when all the nodes are already running in production. The pg_auto_failover monitor then sets a transition for the primary to update its local value of synchronous_standby_names. Live changes, no downtime era. 🦋
🌈 Replication Quorum 🌈
The replication quorum setting is a boolean and defaults to true, and it can be set per-node. Pg_auto_failover includes a given node in synchronous_standby_names only when the replication quorum parameter has been set to true. Which means: asynchronous replication will be used for nodes where replication-quorum is set to false. Noted. 📝
You can also force asynchronous replication globally by setting replication quorum to false on all the nodes in a formation. But remember — failovers will happen, so set your replication settings on the current primary node too when needed, because it's going to be a standby later. Main character today, understudy tomorrow. 👑
To set this parameter to either true or false, use one of these commands:
pg_autoctl set node replication-quorum true
pg_autoctl set node replication-quorum false🌸 Candidate Priority 🌸
The candidate priority setting is an integer that can be set to any value between 0 (zero) and 100 (one hundred). The default value is 50. When the pg_auto_failover monitor decides to orchestrate a failover, it uses each node's candidate priority to pick the new primary node.
If you set a node's candidate priority down to zero, that node will never be selected to be promoted as the new primary when a failover is orchestrated by the monitor. The monitor will instead wait until another node registered is healthy and in a position to be promoted. It's giving patient, honestly.
To set this parameter to the value <n>, use:
pg_autoctl set node candidate-priority <n>When nodes have the same candidate priority, the monitor then picks the standby with the most advanced LSN position published to the monitor. And when more than one node has published the same LSN position, a random one is chosen. RNG moment. 🎲
When the candidate for failover has not published the most advanced LSN position in the WAL, pg_auto_failover orchestrates an intermediate step in the failover mechanism. The candidate fetches the missing WAL bytes from one of the standbys with the most advanced LSN position before being promoted. Postgres allows this operation thanks to cascading replication: any standby can be the upstream node for another standby. Cute. 🔥
Important bestie: it is required at all times that at least two nodes have a non-zero candidate priority in any pg_auto_failover formation. Otherwise no failover is possible. Periodt.
✨ Auditing replication settings ✨
The command pg_autoctl get formation settings (also known as pg_autoctl show settings) can be used to get a summary of all the replication settings currently in effect in a formation. Still using the first diagram on this page, we get this summary:
$ pg_autoctl get formation settings
Context | Name | Setting | Value
----------+---------+---------------------------+-------------------------------------------------------------
formation | default | number_sync_standbys | 1
primary | node_A | synchronous_standby_names | 'ANY 1 (pgautofailover_standby_3, pgautofailover_standby_2)'
node | node_A | replication quorum | true
node | node_B | replication quorum | true
node | node_C | replication quorum | true
node | node_A | candidate priority | 50
node | node_B | candidate priority | 50
node | node_C | candidate priority | 50We can see that the number_sync_standbys has been used to compute the current value of the synchronous_standby_names setting on the primary. 💖
Because all the nodes in that example have the same default candidate priority (50), then pg_auto_failover is using the form ANY 1 with the list of standby nodes that are currently participating in the replication quorum.
The entries in the synchronous_standby_names list are meant to match the application_name connection setting used in the primary_conninfo, and the format used by pg_auto_failover there is the format string "pgautofailover_standby_%d" where %d is replaced by the node id. This lets you keep the same connection string to the primary when the node name is changed (using the command pg_autoctl set metadata --name). Clever, right? 🧚
Here we can see the node id of each registered Postgres node with the following command:
$ pg_autoctl show state
Name | Node | Host:Port | LSN | Reachable | Current State | Assigned State
-------+-------+----------------+-----------+-----------+---------------------+--------------------
node_A | 1 | localhost:5001 | 0/7002310 | yes | primary | primary
node_B | 2 | localhost:5002 | 0/7002310 | yes | secondary | secondary
node_C | 3 | localhost:5003 | 0/7002310 | yes | secondary | secondaryWhen setting pg_auto_failover with per formation number_sync_standby and then per node replication quorum and candidate priority replication settings, those properties are then used to compute the synchronous_standby_names value on the primary node. This value is automatically maintained on the primary by pg_auto_failover, and is updated either when replication settings are changed or when a failover happens. 💫
The other situation when the pg_auto_failover replication settings are used is a candidate election when a failover happens and there is more than two nodes registered in a group. Then the node with the highest candidate priority is selected, as detailed above in the Candidate Priority section.
🌈 Sample architectures with three standby nodes 🌈
When setting the three parameters above, it's possible to design very different Postgres architectures for your production needs.
[image: pg_auto_failover architecture with three standby nodes] [image] pg_auto_failover architecture with three standby nodes.UNINDENT
In this case, the system is set up with three standby nodes all set the same way, with default parameters. The default parameters support setting number_sync_standbys = 2. This means Postgres will maintain three copies of the production data set at all times. Triple protection, we love to see it. 🫶
On the other hand, if two standby nodes were to fail at the same time, despite the fact that two copies of the data are still maintained, the Postgres service would be degraded to read-only. 😭
With this architecture diagram, here's the summary that we obtain:
$ pg_autoctl show settings
Context | Name | Setting | Value
----------+---------+---------------------------+---------------------------------------------------------------------------------------
formation | default | number_sync_standbys | 2
primary | node_A | synchronous_standby_names | 'ANY 2 (pgautofailover_standby_2, pgautofailover_standby_4, pgautofailover_standby_3)'
node | node_A | replication quorum | true
node | node_B | replication quorum | true
node | node_C | replication quorum | true
node | node_D | replication quorum | true
node | node_A | candidate priority | 50
node | node_B | candidate priority | 50
node | node_C | candidate priority | 50
node | node_D | candidate priority | 50🦋 Sample architecture with three standby nodes, one async 🦋
[image: pg_auto_failover architecture with three standby nodes, one async] [image] pg_auto_failover architecture with three standby nodes, one async.UNINDENT
In this case, the system is set up with two standby nodes participating in the replication quorum, allowing for number_sync_standbys = 1. The system always maintains at least two copies of the data set — one on the primary, another on either node B or node C. Whenever we lose one of those nodes, we can hold to the guarantee of having two copies of the data set. 💪
Additionally, we have the standby server D which has been set up to not participate in the replication quorum. Node D will not be found in the synchronous_standby_names list of nodes. Also, node D is set up to never be a candidate for failover, with candidate-priority = 0. Off the roster, bestie. 🙈
This architecture would fit a situation where nodes A, B, and C are deployed in the same data center or availability zone and node D in another one. Those three nodes are set up to support the main production traffic and implement high availability of both the Postgres service and the data set.
Node D might be set up for Business Continuity in case the first data center is lost, or maybe for reporting needs on another application domain.
With this architecture diagram, here's the summary that we obtain:
pg_autoctl show settings
Context | Name | Setting | Value
----------+---------+---------------------------+-------------------------------------------------------------
formation | default | number_sync_standbys | 1
primary | node_A | synchronous_standby_names | 'ANY 1 (pgautofailover_standby_2, pgautofailover_standby_3)'
node | node_A | replication quorum | true
node | node_B | replication quorum | true
node | node_C | replication quorum | true
node | node_D | replication quorum | false
node | node_A | candidate priority | 50
node | node_B | candidate priority | 50
node | node_C | candidate priority | 50
node | node_D | candidate priority | 0🔄 FAILOVER STATE MACHINE 🔄
✨ Introduction ✨
okay bestie, sit down, because this is the part where it all clicks 💖 pg_auto_failover runs on a state machine, and the whole point is that everything happens in a highly controlled way — no vibes-based failover here, no cap.
Here's the tea: keepers tell the monitor about new events (or, uh, they just don't — and the monitor notices that too 🙈). Based on that, the monitor hands every node two things: a current state and a goal state.
- The current state is a strong guarantee of what the node can actually do right now. Periodt.
- States themselves do nothing. They're not the main character — they're the vibe check. All the actual actions happen during state transitions.
- The goal state is the monitor telling the keeper "bestie, try to get here."
🌱 Example of state transitions in a new cluster 🌱
the best way to actually get the states? watch a cluster go from its little baby newborn era 👶 to full high availability. it's giving character development.
So: after you start a monitor and run keeper init for the first data node (node A), the monitor registers that node in the state init with a goal state of single. What does init mean? Literally just "this node exists." The monitor knows nothing else about it, because the keeper isn't running continuously over there reporting health yet. it's a mystery ✨
Once the keeper is actually running and reporting its health to the monitor, the monitor assigns it the state single — meaning it's just a regular old Postgres server, no failover, nothing fancy. And because there aren't other nodes in the cluster yet, the monitor also gives node A the goal state of single — there's literally nothing for that keeper to change. chill.
The moment a new node (node B) gets initialized, the monitor hands node A the goal state wait_primary. Still no failover yet, but now there's hope — a secondary might sync with it soon 🥹. To make the transition from single to wait_primary, node A's keeper adds node B's hostname to pg_hba.conf so a hot standby replication connection is allowed. gateway opened, bestie.
Meanwhile node B transitions into wait_standby, with the goal of... initially just staying in wait_standby. It can't do anything but wait until node A gives it access to connect. patience is a virtue, babe 🧘 Once node A has made it to wait_primary, the monitor gives B the goal of catchingup — green light 🟢 — so B's keeper can go from wait_standby to catchingup. That transition is the whole run pg_basebackup, edit recovery.conf, and restart PostgreSQL in Hot Standby node.
node B reports to the monitor when it's in hot standby mode and able to connect to node A. Then the monitor gives node B the goal state secondary and A the goal of primary. Postgres ships WAL logs from node A and replays them on B. Eventually B is all caught up and tells the monitor — specifically, B reports its pg_stat_replication.sync_state and WAL replay lag. And at that glorious moment the monitor assigns A the state primary (goal: primary) and B secondary (goal: secondary). ate. 💅
📊 State reference 📊
here's the diagram of the whole pg_auto_failover State Machine. it's missing the links to the single state, which can always be reached when you remove all the other nodes. iykyk.
[image: pg_auto_failover Finite State Machine diagram] [image]
pg_auto_failover Finite State Machine diagram.UNINDENTso in that diagram you can see we've got a list of six states where the application can connect to a read-write Postgres service: single, wait_primary, primary, prepare_maintenance, and apply_settings.
🐣 Init 🐣
A node gets the init state the first time it's registered with the monitor. At this point we know nothing about it beyond "it exists" 💀. If no other node has been registered with the monitor for the same formation and group ID, then this node gets the goal state single. Otherwise? Goal state wait_standby. that's it, that's the whole rule.
1️⃣ Single 1️⃣
There's only one node in the group. It behaves like a totally regular PostgreSQL instance — no high availability, no failover, just vibing solo. If the administrator removes a node, the other node will revert to the single state. 🌸
🕰️ Wait_primary 🕰️
This one applies to a node that's intended to be the primary but isn't there yet. The primary-to-be at this point knows the secondary's node name or IP address, and it has granted that node hot standby access in the pg_hba.conf file. almost there bestie 🫶
Now, the wait_primary state can be caused by two very different things:
- a new potential secondary being registered with the monitor — that's the good one ✨
- an existing secondary becoming unhealthy — yeah that's the bad one 😭
In that second case, during the transition from primary to wait_primary, the primary node's keeper disables synchronous replication on the node. It also cancels currently blocked queries. so, heads up.
🚪 Join_primary 🚪
Applied to a primary node when another standby is joining the group. This lets the primary node apply the necessary changes to its HBA setup before allowing the new node joining the system to run the pg_basebackup command.
IMPORTANT: This state has been deprecated, and is no longer assigned to nodes. Any time we would have usedjoin_primarybefore, we now useprimaryinstead. it's retired, she's resting 👑
👑 Primary 👑
A healthy secondary node exists and has caught up with WAL replication. Specifically — and this is the precise bit — the keeper reports the primary state only when it has verified that the secondary is reported "sync" in pg_stat_replication.sync_state, and with a WAL lag of 0.
The primary state is a strong assurance. It's the only state where we know we can fail over when required. that's the whole point, bestie.
Also, during the transition from wait_primary to primary, the keeper also enables synchronous replication. Which means that after a failover the secondary will be fully up to date. slay.
⌛ Wait_standby ⌛
The monitor decides this node is a standby. The node must wait until the primary has authorized it to connect and set up hot standby replication. no notes, just waiting 🧚
🏃 Catchingup 🏃
The monitor assigns catchingup to the standby node when the primary is ready for a replication connection (pg_hba.conf has been properly edited, connection role added, etc).
The standby node's keeper runs pg_basebackup, connecting to the primary's hostname and port. The keeper then edits recovery.conf and starts PostgreSQL in hot standby node. let's gooo 🔥
💫 Secondary 💫
A node with this state is acting as a hot standby for the primary, and is up to date with the WAL log there. Specifically, it is within 16MB or 1 WAL segment of the primary. yes, those are the exact numbers, don't @ me.
🛠️ Maintenance 🛠️
The cluster administrator can manually move a secondary into the maintenance state to gracefully take it offline. The primary will then transition from state primary to wait_primary, during which time the secondary will be online to accept writes. When the old primary reaches the wait_primary state, then the secondary is safe to take offline with minimal consequences. graceful exit, queen 💅
📦 Prepare_maintenance 📦
The cluster administrator can manually move a primary node into the maintenance state to gracefully take it offline. The primary then transitions to the prepare_maintenance state to make sure the secondary is not missing any writes. In the prepare_maintenance state, the primary shuts down.
🧳 Wait_maintenance 🧳
The custer administrator can manually move a secondary into the maintenance state to gracefully take it offline. But before reaching the maintenance state, we want to switch the primary node to asynchronous replication, in order to avoid writes being blocked. In the state wait_maintenance the standby waits until the primary has reached wait_primary.
🚰 Draining 🚰
A state between primary and demoted where replication buffers finish flushing. A draining node will not accept new client writes, but will continue to send existing data to the secondary.
To implement that with Postgres we actually stop the service. When stopping, Postgres ensures that the current replication buffers are flushed correctly to synchronous standbys. no half-flushed buffers in this house 🚫
📉 Demoted 📉
The primary keeper or its database were unresponsive past a certain threshold. The monitor assigns demoted state to the primary to avoid a split-brain scenario where there might be two nodes that don't communicate with each other and both accept client writes. yeah. two primaries. absolutely not. 💀
In that state the keeper stops PostgreSQL and prevents it from running.
⏱️ Demote_timeout ⏱️
If the monitor assigns the primary a demoted goal state but the primary keeper doesn't acknowledge transitioning to that state within a timeout window, then the monitor assigns demote_timeout to the primary.
Most commonly this may happen when the primary machine goes silent. The keeper is not reporting to the monitor. 📵
🛑 Stop_replication 🛑
The stop_replication state is meant to ensure that the primary goes to the demoted state before the standby goes to single and accepts writes (in case the primary can't contact the monitor anymore). Before promoting the secondary node, the keeper stops PostgreSQL on the primary to avoid split-brain situations.
For safety, when the primary fails to contact the monitor and fails to see the pg_auto_failover connection in pg_stat_replication, then it goes to the demoted state of its own accord. self-aware king behavior 😤
🎬 Prepare_promotion 🎬
The prepare_promotion state is meant to prepare the standby server to being promoted. This state allows synchronisation on the monitor, making sure that the primary has stopped Postgres before promoting the secondary, hence preventing split brain situations. safety first, always 🫶
📍 Report_LSN 📍
The report_lsn state is assigned to standby nodes when a failover is orchestrated and there are several standby nodes. In order to pick the furthest standby in the replication, pg_auto_failover first needs a fresh report of the current LSN position reached on each standby node.
When a node reaches the report_lsn state, the replication stream is stopped, by restarting Postgres without a primary_conninfo. This allows the primary node to detect Network Partitions — i.e. when the primary can't connect to the monitor and there's no standby listed in pg_stat_replication. receipts, basically 🧾
⏩ Fast_forward ⏩
The fast_forward state is assigned to the selected promotion candidate during a failover when it won the election thanks to the candidate priority settings, but the selected node is not the most advanced standby node as reported in the report_lsn state.
Missing WAL bytes are fetched from one of the most advanced standby nodes by using Postgres cascading replication features: it is possible to use any standby node in the primary_conninfo. ✨
💀 Dropped
The Dropped state is assigned to a node when the pg_autoctl drop node command is used. It's giving "we're breaking up but let me get my stuff first" — this lets the node implement specific local actions before being entirely removed from the monitor database. 🫶
When a node reports reaching the Dropped state, the monitor removes its entry. If a node is not reporting anymore — maybe because it's completely unavailable, like, ghosted fr fr — then it's possible to run the pg_autoctl drop node --force command, and then the node entry is removed from the monitor. 💅
🧚 pg_auto_failover keeper's State Machine
When built in TEST mode, it is then possible to use the following command to get a visual representation of the Keeper's Finite State Machine:
$ PG_AUTOCTL_DEBUG=1 pg_autoctl do fsm gv | dot -Tsvg > fsm.svgThe dot program is part of the Graphviz suite and produces the following output:
[image: Keeper state machine] [image] Keeper State Machine.
🔥 FAILOVER AND FAULT TOLERANCE
At the heart of the pg_auto_failover implementation is a State Machine. The state machine is driven by the monitor, and its transitions are implemented in the keeper service, which then reports success to the monitor. main character energy tbh
The keeper is allowed to retry transitions as many times as needed until they succeed, and reports also failures to reach the assigned state to the monitor node. The monitor also implements frequent health-checks targeting the registered PostgreSQL nodes. 👑
When the monitor detects something is not as expected, it takes action by assigning a new goal state to the keeper, that is responsible for implementing the transition to this new state, and then reporting.
🩺 Unhealthy Nodes
The pg_auto_failover monitor is responsible for running regular health-checks with every PostgreSQL node it manages. A health-check is successful when it is able to connect to the PostgreSQL node using the PostgreSQL protocol (libpq), imitating the pg_isready command.
How frequent those health checks are (5s by default), the PostgreSQL connection timeout in use (5s by default), and how many times to retry in case of a failure before marking the node unhealthy (2 by default) are GUC variables that you can set on the Monitor node itself. Remember, the monitor is implemented as a PostgreSQL extension, so the setup is a set of PostgreSQL configuration settings:
SELECT name, setting
FROM pg_settings
WHERE name ~ 'pgautofailover\.health';
name | setting
-----------------------------------------+---------
pgautofailover.health_check_max_retries | 2
pgautofailover.health_check_period | 5000
pgautofailover.health_check_retry_delay | 2000
pgautofailover.health_check_timeout | 5000
(4 rows)The pg_auto_failover keeper also reports if PostgreSQL is running as expected. This is useful for situations where the PostgreSQL server / OS is running fine and the keeper (pg_autoctl run) is still active, but PostgreSQL has failed. Situations might include File System is Full on the WAL disk, some file system level corruption, missing files, etc. 😭
Here's what happens to your PostgreSQL service in case of any single-node failure is observed:
- Primary node is monitored unhealthy 💀
When the primary node is unhealthy, and only when the secondary node is itself in good health, then the primary node is asked to transition to the DRAINING state, and the attached secondary is asked to transition to the state PREPARE_PROMOTION. In this state, the secondary is asked to catch-up with the WAL traffic from the primary, and then report success.
The monitor then continues orchestrating the promotion of the standby: it stops the primary (implementing STONITH in order to prevent any data loss), and promotes the secondary into being a primary now. ✨
Depending on the exact situation that triggered the primary unhealthy, it's possible that the secondary fails to catch-up with WAL from it, in that case after the PREPARE_PROMOTION_CATCHUP_TIMEOUT the standby reports success anyway, and the failover sequence continues from the monitor.
- Secondary node is monitored unhealthy 🙈
When the secondary node is unhealthy, the monitor assigns to it the state CATCHINGUP, and assigns the state WAIT_PRIMARY to the primary node. When implementing the transition from PRIMARY to WAIT_PRIMARY, the keeper disables synchronous replication.
When the keeper reports an acceptable WAL difference in the two nodes again, then the replication is upgraded back to being synchronous. While a secondary node is not in the SECONDARY state, secondary promotion is disabled.
- Monitor node has failed 😭
Then the primary and secondary node just work as if you didn't have setup pg_auto_failover in the first place, as the keeper fails to report local state from the nodes. Also, health checks are not performed. It means that no automated failover may happen, even if needed.
🌐 Network Partitions
Adding to those simple situations, pg_auto_failover is also resilient to Network Partitions. Here's the list of situation that have an impact to pg_auto_failover behavior, and the actions taken to ensure High Availability of your PostgreSQL service:
- Primary can't connect to Monitor 🫠
Then it could be that either the primary is alone on its side of a network split, or that the monitor has failed. The keeper decides depending on whether the secondary node is still connected to the replication slot, and if we have a secondary, continues to serve PostgreSQL queries.
Otherwise, when the secondary isn't connected, and after the NETWORK_PARTITION_TIMEOUT has elapsed, the primary considers it might be alone in a network partition: that's a potential split brain situation and with only one way to prevent it. The primary stops, and reports a new state of DEMOTE_TIMEOUT.
The network_partition_timeout can be setup in the keeper's configuration and defaults to 20s.
- Monitor can't connect to Primary 💔
Once all the retries have been done and the timeouts are elapsed, then the primary node is considered unhealthy, and the monitor begins the failover routine. This routine has several steps, each of them allows to control our expectations and step back if needed.
For the failover to happen, the secondary node needs to be healthy and caught-up with the primary. Only if we timeout while waiting for the WAL delta to resorb (30s by default) then the secondary can be promoted with uncertainty about the data durability in the group.
- Monitor can't connect to Secondary 🌸
As soon as the secondary is considered unhealthy then the monitor changes the replication setting to asynchronous on the primary, by assigning it the WAIT_PRIMARY state. Also the secondary is assigned the state CATCHINGUP, which means it can't be promoted in case of primary failure.
As the monitor tracks the WAL delta between the two servers, and they both report it independently, the standby is eligible to promotion again as soon as it's caught-up with the primary again, and at this time it is assigned the SECONDARY state, and the replication will be switched back to synchronous. 👑
🧠 Failure handling and network partition detection
If a node cannot communicate to the monitor, either because the monitor is down or because there is a problem with the network, it will simply remain in the same state until the monitor comes back.
If there is a network partition, it might be that the monitor and secondary can still communicate and the monitor decides to promote the secondary since the primary is no longer responsive. Meanwhile, the primary is still up-and-running on the other side of the network partition. If a primary cannot communicate to the monitor it starts checking whether the secondary is still connected. In PostgreSQL, the secondary connection automatically times out after 30 seconds. If last contact with the monitor and the last time a connection from the secondary was observed are both more than 30 seconds in the past, the primary concludes it is on the losing side of a network partition and shuts itself down. It may be that the secondary and the monitor were actually down and the primary was the only node that was alive, but we currently do not have a way to distinguish such a situation. As with consensus algorithms, availability can only be correctly preserved if at least 2 out of 3 nodes are up. 😤
In asymmetric network partitions, the primary might still be able to talk to the secondary, while unable to talk to the monitor. During failover, the monitor therefore assigns the secondary the stop_replication state, which will cause it to disconnect from the primary. After that, the primary is expected to shut down after at least 30 and at most 60 seconds. To factor in worst-case scenarios, the monitor waits for 90 seconds before promoting the secondary to become the new primary. 💅
🔐 SECURITY SETTINGS FOR PG_AUTO_FAILOVER
In order to be able to orchestrate fully automated failovers, pg_auto_failover needs to be able to establish the following Postgres connections:
- from the monitor node to each Postgres node to check the node's “health” 🩺
- from each Postgres node to the monitor to implement our
node_activeprotocol and fetch the current assigned state for this node - from the secondary node to the primary node for Postgres streaming replication.
Postgres Client authentication is controlled by a configuration file: pg_hba.conf. This file contains a list of rules where each rule may allow or reject a connection attempt.
For pg_auto_failover to work as intended, some HBA rules need to be added to each node configuration. You can choose to provision the pg_hba.conf file yourself thanks to pg_autoctl options' --skip-pg-hba, or you can use the following options to control which kind of rules are going to be added for you. ✨
📜 Postgres HBA rules
For your application to be able to connect to the current Postgres primary servers, some application specific HBA rules have to be added to pg_hba.conf. There is no provision for doing that in pg_auto_failover. 🙈
In other words, it is expected that you have to edit pg_hba.conf to open connections for your application needs. Bestie, that one's on you. 💖
✨ The trust security model ✨
Okay bestie, so trust is literally what it says on the tin — it does zero security validation. None. Nada. 💀 It's lowkey the most popular setting for testing deployments though, because it makes it so easy to verify everything works the way it's supposed to before you go putting actual security restrictions in place. Vibes first, locks later. 🫶
To enable a "trust" security model with pg_auto_failover, you use the pg_autoctl option --auth trust when creating nodes:
$ pg_autoctl create monitor --auth trust ...
$ pg_autoctl create postgres --auth trust ...
$ pg_autoctl create postgres --auth trust ...When you use --auth trust, pg_autoctl adds new HBA rules in the monitor and the Postgres nodes to enable connections as seen above. Slay. 💅
🔐 Authentication with passwords 🔐
To set up pg_auto_failover with a password for connections, you can use one of the password-based authentication methods supported by Postgres, such as password or scram-sha-256. And bestie, we highkey recommend the latter, like in this example:
$ pg_autoctl create monitor --auth scram-sha-256 ...Now here's the thing — pg_autoctl does not set the password for you. 🙈 So step one is setting the database user password in the monitor database, and you do that with this command:
$ psql postgres://monitor.host/pg_auto_failover
> alter user autoctl_node password 'h4ckm3';Now that the monitor is ready with our password set for the autoctl_node user, we can use the password in the monitor connection string that we use when creating Postgres nodes. ✨
On the primary node, we can create the Postgres setup as usual, and then set our replication password — the one we'll use if we get demoted and then re-join as a standby:
$ pg_autoctl create postgres \
--auth scram-sha-256 \
... \
--monitor postgres://autoctl_node:h4ckm3@monitor.host/pg_auto_failover
$ pg_autoctl config set replication.password h4ckm3m0r3The second Postgres node is going to be initialized as a secondary, and pg_autoctl then calls pg_basebackup at create time. So we need to have the replication password already set at that point, and we can achieve that the following way:
$ export PGPASSWORD=h4ckm3m0r3
$ pg_autoctl create postgres \
--auth scram-sha-256 \
... \
--monitor postgres://autoctl_node:h4ckm3@monitor.host/pg_auto_failover
$ pg_autoctl config set replication.password h4ckm3m0r3💡 Note that you can use The Password File mechanism as discussed in the Postgres documentation in order to keep your passwords in a separate file, instead of in your main pg_auto_failover configuration file. This also avoids using passwords in the environment and in command lines. iykyk 🧚
🛡️ Encryption of network communications 🛡️
Postgres knows how to use SSL to enable network encryption of all communications — including authentication with passwords, and the whole entire data set when streaming replication is used. Iconic behavior honestly. 👑
To enable SSL on the server you need an SSL certificate. It could be as simple as a self-signed certificate, and pg_autoctl creates such a certificate for you when you use the --ssl-self-signed command line option:
$ pg_autoctl create monitor --ssl-self-signed ... \
--auth scram-sha-256 ... \
--ssl-mode require \
...
$ pg_autoctl create postgres --ssl-self-signed ... \
--auth scram-sha-256 ... \
...
$ pg_autoctl create postgres --ssl-self-signed ... \
--auth scram-sha-256 ... \
...So in that example we set up SSL connections to encrypt the network traffic, and we still have to set up an authentication mechanism exactly as in the previous sections of this document. Here scram-sha-256 has been selected, and the password will be sent over an encrypted channel. Chef's kiss. 😭
When using the --ssl-self-signed option, pg_autoctl creates a self-signed certificate, as per the Postgres documentation at the Creating Certificates page.
The certificate subject CN defaults to the --hostname parameter, which can be given explicitly or computed by pg_autoctl as either your hostname when you have proper DNS resolution, or your current IP address.
⚠️ Real talk: Self-signed certificates provide protection against eavesdropping; this setup does NOT protect against Man-In-The-Middle attacks nor Impersonation attacks. See PostgreSQL documentation page SSL Support for details. Not the moment to be sloppy, bestie. 💀
🎀 Using your own SSL certificates 🎀
In many cases you're gonna want to install certificates provided by your local security department and signed by a trusted Certificate Authority. In that case one solution is to use --skip-pg-hba and do the whole setup yourself. You got this. 🫶
It is still possible to give the certificates to pg_auto_failover and have it handle the Postgres setup for you:
$ pg_autoctl create monitor --ssl-ca-file root.crt \
--ssl-crl-file root.crl \
--server-cert server.crt \
--server-key server.key \
--ssl-mode verify-full \
...
$ pg_autoctl create postgres --ssl-ca-file root.crt \
--server-cert server.crt \
--server-key server.key \
--ssl-mode verify-full \
...
$ pg_autoctl create postgres --ssl-ca-file root.crt \
--server-cert server.crt \
--server-key server.key \
--ssl-mode verify-full \
...The option --ssl-mode can be used to force connection strings used by pg_autoctl to contain your preferred ssl mode. It defaults to require when using --ssl-self-signed and to allow when --no-ssl is used. Here, we set --ssl-mode to verify-full which requires SSL Certificates Authentication, covered next. ✨
The default --ssl-mode when providing your own certificates (signed by your trusted CA) is then verify-full. This setup applies to the client connection where the server identity is going to be checked against the root certificate provided with --ssl-ca-file and the revocation list optionally provided with the --ssl-crl-file. Both those files are used as the respective parameters sslrootcert and sslcrl in pg_autoctl connection strings to both the monitor and the streaming replication primary server.
💜 SSL Certificates Authentication 💜
Given those files, it is then possible to use certificate-based authentication of client connections. For that, it is necessary to prepare client certificates signed by your root certificate private key and using the target user name as its CN, as per Postgres documentation for Certificate Authentication:
The cn (Common Name) attribute of the certificate will be compared to the requested database user name, and if they match the login will be allowed
For enabling the cert authentication method with pg_auto_failover, you need to prepare a Client Certificate for the user postgres and used by pg_autoctl when connecting to the monitor, to place in ~/.postgresql/postgresql.crt along with its key ~/.postgresql/postgresql.key, in the home directory of the user that runs the pg_autoctl service (which defaults to postgres). 🗂️
Then you need to create a user name map as documented in Postgres page User Name Maps so that your certificate can be used to authenticate pg_autoctl users.
The ident map in pg_ident.conf on the pg_auto_failover monitor should then have the following entry, to allow postgres to connect as the autoctl_node user for pg_autoctl operations:
# MAPNAME SYSTEM-USERNAME PG-USERNAME
# pg_autoctl runs as postgres and connects to the monitor autoctl_node user
pgautofailover postgres autoctl_nodeTo enable streaming replication, the pg_ident.conf file on each Postgres node should now allow the postgres user in the client certificate to connect as the pgautofailover_replicator database user:
# MAPNAME SYSTEM-USERNAME PG-USERNAME
# pg_autoctl runs as postgres and connects to the monitor autoctl_node user
pgautofailover postgres pgautofailover_replicatorGiven that user name map, you can then use the cert authentication method. And as with the pg_ident.conf provisioning, it is best to now provision the HBA rules yourself, using the --skip-pg-hba option:
$ pg_autoctl create postgres --skip-pg-hba --ssl-ca-file ...The HBA rule will use the authentication method cert with a map option, and might then look like the following on the monitor:
# allow certificate based authentication to the monitor
hostssl pg_auto_failover autoctl_node 10.0.0.0/8 cert map=pgautofailoverThen your pg_auto_failover nodes on the 10.0.0.0 network are allowed to connect to the monitor with the user autoctl_node used by pg_autoctl, assuming they have a valid and trusted client certificate. Main character energy for those nodes ngl. 🌟
The HBA rule to use on the Postgres nodes to allow for Postgres streaming replication connections looks like the following:
# allow streaming replication for pg_auto_failover nodes
hostssl replication pgautofailover_replicator 10.0.0.0/8 cert map=pgautofailoverBecause the Postgres server runs as the postgres system user, the connection to the primary node can be made with SSL enabled and will then use the client certificates installed in the postgres home directory in ~/.postgresql/postgresql.{key,cert} locations. 💅
🌸 Postgres HBA provisioning 🌸
So, while pg_auto_failover knows how to manage the Postgres HBA rules that are necessary for your stream replication needs and for its monitor protocol, it will not manage the Postgres HBA rules that are needed for your applications. That part? That's on you, bestie. 🙈
If you have your own HBA provisioning solution, you can include the rules needed for pg_auto_failover and then use the --skip-pg-hba option to the pg_autoctl create commands.
🌱 Enable SSL connections on an existing setup 🌱
Whether you upgrade pg_auto_failover from a previous version that did not have support for the SSL features, or when you started with --no-ssl and later change your mind (it happens!! we've all been there 💀), it is possible with pg_auto_failover to add SSL settings on a system that has already been setup without explicit SSL support.
In this section we detail how to upgrade to SSL settings.
Installing Self-Signed certificates on-top of an already existing pg_auto_failover setup is done with one of the following pg_autoctl command variants, depending if you want self-signed certificates or fully verified ssl certificates:
$ pg_autoctl enable ssl --ssl-self-signed --ssl-mode required
$ pg_autoctl enable ssl --ssl-ca-file root.crt \
--ssl-crl-file root.crl \
--server-cert server.crt \
--server-key server.key \
--ssl-mode verify-fullokay so the pg_autoctl enable ssl command? it ✨edits✨ the postgresql-auto-failover.conf Postgres config file so it matches whatever you just typed on the command line, turns SSL on like you asked, and then goes and updates the pg_autoctl configuration too. multitasking queen 💅
and bestie, the connection string it uses to reach the monitor gets auto-updated by pg_autoctl enable ssl as well. wanna double-check your shiny new config? run this:
$ pg_autoctl config get pg_autoctl.monitorngl here's a thing to watch out for: if you already have a pg_autoctl daemon running, it's gonna try to reload its config once pg_autoctl enable ssl finishes. but sometimes a reload just isn't possible without a restart. so pls pls pls go check the logs of your running daemon to confirm the reload actually worked. if it didn't, you might have to restart the daemon so it picks up the new connection string. no cap 🫶
also the HBA settings? those are not edited, no matter whether you used --skip-pg-hba back at creation time. why? because the host records in the Postgres HBA file match either SSL or non-SSL connection attempts, so your pre-existing setup just keeps on working. if you wanna go the extra mile and beef up your SSL setup, you can manually edit the HBA files and swap those existing lines from host to hostssl, which disallows unencrypted connections on the server side. 🔒
so in summary, upgrading an existing pg_auto_failover setup to enable SSL looks like this:
- run the
pg_autoctl enable sslcommand on your monitor and then on all the Postgres nodes, - on the Postgres nodes, review your
pg_autoctllogs to make sure the reload operation actually took effect, and review your Postgres settings to verify you got the result you expected, - review your HBA rules setup to change the
pg_auto_failoverrules fromhosttohostsslto disallow insecure connections.
🧡 CITUS SUPPORT 🧡
the usual pg_autoctl commands work with Postgres standalone nodes and with Citus nodes. one command set, two vibes, we love versatility 💖
pg_auto_failover architecture with a Citus formation. pg_auto_failover architecture with a Citus formation.
so when you're running pg_auto_failover with Citus, a pg_auto_failover formation is made up of a coordinator plus a bunch of worker nodes.
and when High-Availability is enabled at the formation level — which btw is the default, bestie — you need a minimum of two coordinator nodes: a primary and a secondary coordinator, so a failover can actually be orchestrated when the moment calls for it. 🎀
same energy for the worker nodes: when you use pg_auto_failover for Citus HA, every worker node is its own pg_auto_failover group in the formation, and each worker group is set up with at least two nodes (primary, secondary). iconic structure honestly 👑
🌈 Setting-up your first Citus formation
go peek at our documentation of Citus Tutorial for all the details, it's a full tutorial setup on a single VM, perfect for testing and QA. 🧪
🦋 Citus specific commands and operations
when you're setting up Citus with pg_auto_failover, here are the Citus-specific commands you get. every other pg_autoctl command works as usual when you deploy a Citus formation, so you can use the rest of this documentation to operate your Citus deployments. 💅
🌸 pg_autoctl create coordinator
this one creates a Citus coordinator — aka a Postgres node with the Citus extension loaded and ready to serve as a coordinator. and the coordinator is always placed in the pg_auto_failover group zero of a given formation. always. no exceptions. group zero is its home 🏠
see pg_autoctl create coordinator for details.
IMPORTANT: the default--dbnameis the same as the current system user name, which in many cases is gonna bepostgres. pls pls make sure you use the--dbnameoption with the actual database you're going to use with your application. 🙈 and heads up: Citus does not support multiple databases, so you have to use the database where Citus is created. when you're using Citus, that is essential to the well-behaving of worker failover. fr fr, don't skip this one.
🌟 pg_autoctl create worker
this command creates a new Citus worker node — a Postgres node with the Citus extensions loaded, registered to the Citus coordinator you made with the previous command. and because the Citus coordinator is always given group zero, the pg_auto_failover monitor knows exactly how to reach the Citus coordinator and automate the workers registration. smart, right? 😌
the default worker creation policy is: assign the primary role to the first worker registered, then secondary in the same group, then primary in a new group, etc. round and round, slay ✨
if you wanna extend an existing group to have a third worker node in the same group — unlocking multiple-standby capabilities in your setup — then make sure to use the --group option on the pg_autoctl create worker command. iykyk 🫶
see pg_autoctl create worker for details.
💜 pg_autoctl activate
this command calls the Citus "activation" API, so that a node can be used to host shards for your reference and distributed tables. 👑
when you create a Citus worker, pg_autoctl create worker automatically activates the worker node to the coordinator. so you only need this command when something unexpected happened and you wanna manually make sure the worker node has been activated at the Citus coordinator level. that's your emergency backup plan 🚨
starting with Citus 10, it's also possible to activate the coordinator itself as a node with shard placement. just use pg_autoctl activate on your Citus coordinator node manually to use that feature. ✨
and when the Citus coordinator is activated, there's an extra step needed for it to host shards of distributed tables. if you want your coordinator to have shards, then go have a look at the Citus API ‐ citus_set_node_property to set the shouldhaveshards property to true. 🌈
see pg_autoctl activate for details.
🎀 Citus worker failover
when a failover is orchestrated by pg_auto_failover for a Citus worker node, Citus gives us the opportunity to make that failover close to transparent to the application. how? because the application connects to the coordinator, which in turn connects to the worker nodes, Citus can pause the SQL write traffic on the coordinator for the shards hosted on the failed worker node. then the Postgres failover happens while the traffic stays on the coordinator, and it resumes the second a secondary worker node is ready to accept read-write queries. literally seamless 😭💖
this is implemented thanks to Citus's smart locking strategy in its citus_update_node API, and pg_auto_failover takes full benefit of it with a special built set of FSM transitions made for Citus workers. 🔥
🍓 Citus Secondaries and read-replica
it's possible to set up Citus read-only replicas. this Citus feature lets you use a set of dedicated nodes (both coordinator and workers) to serve read-only traffic — think reporting, analytics, or any other part of your workload that's read-only. 📊
Citus read-replica nodes are a solution for load-balancing. those nodes can't be used as HA failover targets, and so they have their candidate-priority set to zero. and this setting of a read-replica can not be changed later. periodt. 🔒
this setup is done by setting the Citus property citus.use_secondary_nodes to always (it defaults to never), and the Citus property citus.cluster_name to your read-only cluster name.
both of those settings are entirely supported and managed by pg_autoctl when you use the --citus-secondary --cluster-name cluster_d options on the pg_autoctl create coordinator|worker commands. 💅
here's an example where we made a formation with three nodes for HA for the coordinator (one primary and two secondary nodes), then a single worker node with the same three-nodes setup for HA, and then one read-replica environment on top of that, for a total of 8 nodes: 👑
$ pg_autoctl show state
Name | Node | Host:Port | LSN | Reachable | Current State | Assigned State
---------+-------+----------------+-----------+-----------+---------------------+--------------------
coord0a | 0/1 | localhost:5501 | 0/5003298 | yes | primary | primary
coord0b | 0/3 | localhost:5502 | 0/5003298 | yes | secondary | secondary
coord0c | 0/6 | localhost:5503 | 0/5003298 | yes | secondary | secondary
coord0d | 0/7 | localhost:5504 | 0/5003298 | yes | secondary | secondary
worker1a | 1/2 | localhost:5505 | 0/4000170 | yes | primary | primary
worker1b | 1/4 | localhost:5506 | 0/4000170 | yes | secondary | secondary
worker1c | 1/5 | localhost:5507 | 0/4000170 | yes | secondary | secondary
reader1d | 1/8 | localhost:5508 | 0/4000170 | yes | secondary | secondarythe nodes named coord0d and reader1d are members of the read-replica cluster cluster_d. and you can see that a read-replica cluster needs a dedicated coordinator, and then one dedicated worker node per group. 🫶
TIP: it's possible to name the nodes in apg_auto_failoverformation either at creation time or later, using one of those commands:bash $ pg_autoctl create worker --name ... $ pg_autoctl set node metadata --name ...herecoord0dis the node name for the dedicated coordinator forcluster_d, andreader1dis the node name for the dedicated worker forcluster_din the worker group1(the only worker group in that setup). 💖
now, the usual command to show the connection strings for your application lists the read-replica setup like this:
$ pg_autoctl show uri
Type | Name | Connection String
---------+-----------+-------------------------------
monitor | monitor | postgres://autoctl_node@localhost:5500/pg_auto_failover?sslmode=prefer
formation | default | postgres://localhost:5503,localhost:5501,localhost:5502/postgres?target_session_attrs=read-write&sslmode=preferread-replica | cluster_d | postgres://localhost:5504/postgres?sslmode=prefergiven that setup, your application can now use the formation default Postgres URI to connect to the highly-available read-write service, or connect to the read-replica cluster_d service to hit the read-only replica where you can offload some of your SQL workload. 🌈✨
and when you connect to the cluster_d connection string, the Citus properties citus.use_secondary_nodes and citus.cluster_name are automatically set up to their expected values, of course. no manual fiddling required, bestie 💅
💖 CITUS TUTORIAL 💖
okay bestie, buckle up, this one is a whole journey 🦋 In this guide we're gonna build a Citus cluster with a coordinator node and three workers. Every single node gets a secondary for failover, because we do NOT do single points of failure in this house. Then we're gonna simulate failure in the coordinator and the worker nodes and watch the system keep functioning like nothing happened. main character energy fr fr 👑
This tutorial uses docker compose so we can separate the architecture design from all the messy implementation details. That way we get to reason at the architecture level inside this tutorial, and actually see which software component needs to be deployed and run on which node. It's giving "big picture first, plumbing later" ✨
Heads up bestie: the setup in this tutorial is great for replaying at home in the lab. It is NOT intended to be production ready though. Specifically, zero attention has been spent on volume management. After all, this is a tutorial: the goal is to walk through the first steps of using pg_auto_failover to provide HA to a Citus formation. 💅
🌸 Pre-requisites 🌸
When using docker compose we describe a list of services, each service may run on one or more nodes, and each service just runs a single isolated process in a container.
Within the context of a tutorial, or even a development environment, this matches very well to provisioning separate physical machines on-prem, or Virtual Machines either on-prem or in a Cloud service. 🎀
The docker image used in this tutorial is named pg_auto_failover:citus. It can be built locally when using the attached Dockerfile found within the GitHub repository for pg_auto_failover.
To build the image, either use the provided Makefile and run make build, or run the docker build command directly:
$ git clone https://github.com/citusdata/pg_auto_failover
$ cd pg_auto_failover/docs/cluster
$ docker build -t pg_auto_failover:citus -f Dockerfile ../..
$ docker compose build👑 Our first Citus Cluster 👑
To create a cluster we use the following docker compose definition:
version: "3.9" # optional since v1.27.0
services:
monitor:
image: pg_auto_failover:citus
environment:
PGDATA: /tmp/pgaf
command: |
pg_autoctl create monitor --ssl-self-signed --auth trust --run
expose:
- 5432
coord:
image: pg_auto_failover:citus
environment:
PGDATA: /tmp/pgaf
PGUSER: citus
PGDATABASE: citus
PG_AUTOCTL_MONITOR: "postgresql://autoctl_node@monitor/pg_auto_failover"
expose:
- 5432
command: |
pg_autoctl create coordinator --ssl-self-signed --auth trust --pg-hba-lan --run
worker:
image: pg_auto_failover:citus
environment:
PGDATA: /tmp/pgaf
PGUSER: citus
PGDATABASE: citus
PG_AUTOCTL_MONITOR: "postgresql://autoctl_node@monitor/pg_auto_failover"
expose:
- 5432
command: |
pg_autoctl create worker --ssl-self-signed --auth trust --pg-hba-lan --runTo run the full Citus cluster with HA from this definition, we can use the following command:
$ docker compose up --scale coord=2 --scale worker=6The command above starts the services up. The command also specifies a --scale option that is different for each service. We need:
- one monitor node, and the default scale for a service is
1, - one primary Citus coordinator node and one secondary
Cituscoordinatornode, which is to say two coordinator nodes, - and three Citus worker nodes, each worker with both a primary Postgres node and a secondary Postgres node, so that's a scale of
6here.
The default policy for the pg_auto_failover monitor is to assign a primary and a secondary per auto failover Group. In our case, every node being provisioned with the same command, we benefit from that default policy:
$ pg_autoctl create worker --ssl-self-signed --auth trust --pg-hba-lan --runWhen provisioning a production cluster, it is often required to have a better control over which node participates in which group, then using the --group N option in the pg_autoctl create worker command line.
Within a given group, the first node that registers is a primary, and the other nodes are secondary nodes. The monitor takes care of that in a way that we don't have to. In a High Availability setup, every node should be ready to be promoted primary at any time, so knowing which node in a group is assigned primary first is not very interesting. 😌
While the cluster is being provisionned by docker compose, you can run the following command and have a dynamic dashboard to follow what's happening. The following command is like top for pg_auto_failover:
$ docker compose exec monitor pg_autoctl watchBecause the pg_basebackup operation that is used to create the secondary nodes takes some time when using Citus, because of the first CHECKPOINT which is quite slow. So at first when inquiring about the cluster state you might see the following output:
$ docker compose exec monitor pg_autoctl show state
Name | Node | Host:Port | TLI: LSN | Connection | Reported State | Assigned State
---------+-------+-------------------+----------------+--------------+---------------------+--------------------
coord0a | 0/1 | cd52db444544:5432 | 1: 0/200C4A0 | read-write | wait_primary | wait_primary
coord0b | 0/2 | 66a31034f2e4:5432 | 1: 0/0 | none ! | wait_standby | catchingup
worker1a | 1/3 | dae7c062e2c1:5432 | 1: 0/2003B18 | read-write | wait_primary | wait_primary
worker1b | 1/4 | 397e6069b09b:5432 | 1: 0/0 | none ! | wait_standby | catchingup
worker2a | 2/5 | 5bf86f9ef784:5432 | 1: 0/2006AB0 | read-write | wait_primary | wait_primary
worker2b | 2/6 | 23498b801a61:5432 | 1: 0/0 | none ! | wait_standby | catchingup
worker3a | 3/7 | c23610380024:5432 | 1: 0/2003B18 | read-write | wait_primary | wait_primary
worker3b | 3/8 | 2ea8aac8a04a:5432 | 1: 0/0 | none ! | wait_standby | catchingupAfter a while though (typically around a minute or less), you can run that same command again for stable result:
$ docker compose exec monitor pg_autoctl show state
Name | Node | Host:Port | TLI: LSN | Connection | Reported State | Assigned State
---------+-------+-------------------+----------------+--------------+---------------------+--------------------
coord0a | 0/1 | cd52db444544:5432 | 1: 0/3127AD0 | read-write | primary | primary
coord0b | 0/2 | 66a31034f2e4:5432 | 1: 0/3127AD0 | read-only | secondary | secondary
worker1a | 1/3 | dae7c062e2c1:5432 | 1: 0/311B610 | read-write | primary | primary
worker1b | 1/4 | 397e6069b09b:5432 | 1: 0/311B610 | read-only | secondary | secondary
worker2a | 2/5 | 5bf86f9ef784:5432 | 1: 0/311B610 | read-write | primary | primary
worker2b | 2/6 | 23498b801a61:5432 | 1: 0/311B610 | read-only | secondary | secondary
worker3a | 3/7 | c23610380024:5432 | 1: 0/311B648 | read-write | primary | primary
worker3b | 3/8 | 2ea8aac8a04a:5432 | 1: 0/311B648 | read-only | secondary | secondaryYou can see from the above that the coordinator node has a primary and a secondary instance for high availability. When connecting to the coordinator, clients should try connecting to whichever instance is running and supports reads and writes. 🫶
We can review the available Postgres URIs with the pg_autoctl show uri command:
$ docker compose exec monitor pg_autoctl show uri
Type | Name | Connection String
-------------+---------+-------------------------------
monitor | monitor | postgres://autoctl_node@552dd89d5d63:5432/pg_auto_failover?sslmode=require
formation | default | postgres://66a31034f2e4:5432,cd52db444544:5432/citus?target_session_attrs=read-write&sslmode=requireTo check that Citus worker nodes have been registered to the coordinator, we can run a psql session right from the coordinator container:
$ docker compose exec coord psql -d citus -c 'select * from citus_get_active_worker_nodes();'
node_name | node_port
--------------+-----------
dae7c062e2c1 | 5432
5bf86f9ef784 | 5432
c23610380024 | 5432
(3 rows)We are now reaching the limits of using a simplified docker compose setup. When using the --scale option, it is not possible to give a specific hostname to each running node, and then we get a randomly generated string instead or useful node names such as worker1a or worker3b. 💀
🌈 Create a Citus Cluster, take two 🌈
In order to implement the following architecture, we need to introduce a more complex docker compose file than in the previous section.
🎀 [image: pg_auto_failover architecture with a Citus formation] — pg_auto_failover architecture with a Citus formation.
This time we create a cluster using the following docker compose definition:
x-coord: &coordinator
image: pg_auto_failover:citus
environment:
PGDATA: /tmp/pgaf
PGUSER: citus
PGDATABASE: citus
PG_AUTOCTL_HBA_LAN: true
PG_AUTOCTL_AUTH_METHOD: "trust"
PG_AUTOCTL_SSL_SELF_SIGNED: true
PG_AUTOCTL_MONITOR: "postgresql://autoctl_node@monitor/pg_auto_failover"
expose:
- 5432
x-worker: &worker
image: pg_auto_failover:citus
environment:
PGDATA: /tmp/pgaf
PGUSER: citus
PGDATABASE: citus
PG_AUTOCTL_HBA_LAN: true
PG_AUTOCTL_AUTH_METHOD: "trust"
PG_AUTOCTL_SSL_SELF_SIGNED: true
PG_AUTOCTL_MONITOR: "postgresql://autoctl_node@monitor/pg_auto_failover"
expose:
- 5432
services:
app:
build:
context: .
dockerfile: Dockerfile.app
environment:
PGUSER: citus
PGDATABASE: citus
PGHOST: coord0a,coord0b
PGPORT: 5432
PGAPPNAME: demo
PGSSLMODE: require
PGTARGETSESSIONATTRS: read-write
monitor:
image: pg_auto_failover:citus
environment:
PGDATA: /tmp/pgaf
PG_AUTOCTL_SSL_SELF_SIGNED: true
expose:
- 5432
command: |
pg_autoctl create monitor --auth trust --run coord0a:
<<: *coordinator
hostname: coord0a
command: |
pg_autoctl create coordinator --name coord0a --run
coord0b:
<<: *coordinator
hostname: coord0b
command: |
pg_autoctl create coordinator --name coord0b --run
worker1a:
<<: *worker
hostname: worker1a
command: |
pg_autoctl create worker --group 1 --name worker1a --run
worker1b:
<<: *worker
hostname: worker1b
command: |
pg_autoctl create worker --group 1 --name worker1b --run
worker2a:
<<: *worker
hostname: worker2a
command: |
pg_autoctl create worker --group 2 --name worker2a --run
worker2b:
<<: *worker
hostname: worker2b
command: |
pg_autoctl create worker --group 2 --name worker2b --run
worker3a:
<<: *worker
hostname: worker3a
command: |
pg_autoctl create worker --group 3 --name worker3a --run
worker3b:
<<: *worker
hostname: worker3b
command: |
pg_autoctl create worker --group 3 --name worker3b --runokay so bestie, this one is a little more extra than the last definition and honestly? it ate. 💅 we're using YAML anchors and aliases to make a template for our coordinator nodes and our worker nodes, and then we just slap that template onto the actual nodes. iconic, no notes.
and ALSO — this time we spin up an application service (she's named app) that just chills in the background so we can connect to our current primary coordinator later. peep Dockerfile.app for the whole entire definition of that service, iykyk. 🦋
We start this cluster with a glow-up of a command line this time, way shorter:
$ docker compose upand the result? this cluster. look at her:
$ docker compose exec monitor pg_autoctl show state
Name | Node | Host:Port | TLI: LSN | Connection | Reported State | Assigned State
---------+-------+---------------+----------------+--------------+---------------------+--------------------
coord0a | 0/3 | coord0a:5432 | 1: 0/312B040 | read-write | primary | primary
coord0b | 0/4 | coord0b:5432 | 1: 0/312B040 | read-only | secondary | secondary
worker1a | 1/1 | worker1a:5432 | 1: 0/311C550 | read-write | primary | primary
worker1b | 1/2 | worker1b:5432 | 1: 0/311C550 | read-only | secondary | secondary
worker2b | 2/7 | worker2b:5432 | 2: 0/5032698 | read-write | primary | primary
worker2a | 2/8 | worker2a:5432 | 2: 0/5032698 | read-only | secondary | secondary
worker3a | 3/5 | worker3a:5432 | 1: 0/311C870 | read-write | primary | primary
worker3b | 3/6 | worker3b:5432 | 1: 0/311C870 | read-only | secondary | secondaryand then we've got this application connection string to actually use, periodt:
$ docker compose exec monitor pg_autoctl show uri
Type | Name | Connection String
-------------+---------+-------------------------------
monitor | monitor | postgres://autoctl_node@f0135b83edcd:5432/pg_auto_failover?sslmode=require
formation | default | postgres://coord0b:5432,coord0a:5432/citus?target_session_attrs=read-write&sslmode=requireand finally, the nodes being registered as Citus worker nodes starts making sooo much more sense now:
$ docker compose exec coord0a psql -d citus -c 'select * from citus_get_active_worker_nodes()'
node_name | node_port
-----------+-----------
worker1a | 5432
worker3a | 5432
worker2b | 5432
(3 rows)IMPORTANT: 🚨 bestie listen. at this point it's super important to clock that the Citus coordinator only knows about the primary nodes in each group. the High Availability mechanisms are all implemented in pg_auto_failover, which mostly uses the Citus API citus_update_node during worker node failovers. that's the whole vibe. don't forget it. 💜
🎀 Our first Citus worker failover 🎀
so we look at that citus_get_active_worker_nodes() output and we've got worker1a, worker2b, and worker3a. and like, as we mentioned before, that should have zero impact on the operations of the Citus cluster when the nodes are all dimensionned the same. it's giving balanced. it's giving fine. 🙈
that said, some of you readers are gonna want the A nodes as primaries to get started with, and honestly? valid. so let's implement our first worker failover then. with pg_auto_failover this is as easy as doing:
$ docker compose exec monitor pg_autoctl perform failover --group 2
15:40:03 9246 INFO Waiting 60 secs for a notification with state "primary" in formation "default" and group 2
15:40:03 9246 INFO Listening monitor notifications about state changes in formation "default" and group 2
15:40:03 9246 INFO Following table displays times when notifications are received
Time | Name | Node | Host:Port | Current State | Assigned State
---------+----------+-------+---------------+---------------------+--------------------
22:58:42 | worker2b | 2/7 | worker2b:5432 | primary | draining
22:58:42 | worker2a | 2/8 | worker2a:5432 | secondary | prepare_promotion
22:58:42 | worker2a | 2/8 | worker2a:5432 | prepare_promotion | prepare_promotion
22:58:42 | worker2a | 2/8 | worker2a:5432 | prepare_promotion | wait_primary
22:58:42 | worker2b | 2/7 | worker2b:5432 | primary | demoted
22:58:42 | worker2b | 2/7 | worker2b:5432 | draining | demoted
22:58:42 | worker2b | 2/7 | worker2b:5432 | demoted | demoted
22:58:43 | worker2a | 2/8 | worker2a:5432 | wait_primary | wait_primary
22:58:44 | worker2b | 2/7 | worker2b:5432 | demoted | catchingup
22:58:46 | worker2b | 2/7 | worker2b:5432 | catchingup | catchingup
22:58:46 | worker2b | 2/7 | worker2b:5432 | catchingup | secondary
22:58:46 | worker2b | 2/7 | worker2b:5432 | secondary | secondary
22:58:46 | worker2a | 2/8 | worker2a:5432 | wait_primary | primary
22:58:46 | worker2a | 2/8 | worker2a:5432 | primary | primaryso it took like, around 5 seconds to do a full worker failover in worker group 2. speed. ⚡ now we'll do the same on the group 1 to fix the other situation, and review the resulting cluster state:
$ docker compose exec monitor pg_autoctl show state
Name | Node | Host:Port | TLI: LSN | Connection | Reported State | Assigned State
---------+-------+---------------+----------------+--------------+---------------------+--------------------
coord0a | 0/3 | coord0a:5432 | 1: 0/312ADA8 | read-write | primary | primary
coord0b | 0/4 | coord0b:5432 | 1: 0/312ADA8 | read-only | secondary | secondary
worker1a | 1/1 | worker1a:5432 | 1: 0/311B610 | read-write | primary | primary
worker1b | 1/2 | worker1b:5432 | 1: 0/311B610 | read-only | secondary | secondary
worker2b | 2/7 | worker2b:5432 | 2: 0/50000D8 | read-only | secondary | secondary
worker2a | 2/8 | worker2a:5432 | 2: 0/50000D8 | read-write | primary | primary
worker3a | 3/5 | worker3a:5432 | 1: 0/311B648 | read-write | primary | primary
worker3b | 3/6 | worker3b:5432 | 1: 0/311B648 | read-only | secondary | secondarywhich, seen from the Citus coordinator, looks like the following:
$ docker compose exec coord0a psql -d citus -c 'select * from citus_get_active_worker_nodes()'
node_name | node_port
-----------+-----------
worker1a | 5432
worker3a | 5432
worker2a | 5432
(3 rows)🌈 Distribute Data to Workers 🌈
let's create a database schema with a single distributed table. single. one. table. she's minimal, she's powerful:
$ docker compose exec app psql
-- in psql
CREATE TABLE companies
(
id bigserial PRIMARY KEY,
name text NOT NULL,
image_url text,
created_at timestamp without time zone NOT NULL,
updated_at timestamp without time zone NOT NULL
);
SELECT create_distributed_table('companies', 'id');next, download and ingest some sample data — still from within our psql session, bestie, we're not going anywhere:
\copy companies from program 'curl -o- https://examples.citusdata.com/mt_ref_arch/companies.csv' with csv
# ( COPY 75 )💀 Handle Worker Failure 💀
now we're gonna intentionally crash a worker's primary node and watch how the pg_auto_failover monitor unregisters that node in the coordinator and registers the secondary instead. chaos, but make it controlled. 😭
# the pg_auto_failover keeper process will be unable to resurrect
# the worker node if pg_control has been removed
$ docker compose exec worker1a rm /tmp/pgaf/global/pg_control
# shut it down
$ docker compose exec worker1a /usr/lib/postgresql/14/bin/pg_ctl stop -D /tmp/pgafthe keeper will attempt to start worker 1a three times and then report the failure to the monitor, who promotes worker1b to replace worker1a. Citus worker worker1a is unregistered with the coordinator node, and worker1b is registered in its stead. she stepped up. she took the crown. 👑
asking the coordinator for active worker nodes now shows worker1b, worker2a, and worker3a:
$ docker compose exec app psql -c 'select * from master_get_active_worker_nodes();'
node_name | node_port
-----------+-----------
worker3a | 5432
worker2a | 5432
worker1b | 5432
(3 rows)finally, verify that all rows of data are still present. all of them. every single one:
$ docker compose exec app psql -c 'select count(*) from companies;'
count
-------
75meanwhile, the keeper on worker 1a heals the node. she runs pg_basebackup to fetch the current PGDATA from worker1a. this restores, among other things, a brand new copy of the file we just removed (rip pg_control, you were so real 🙏). after streaming replication completes, worker1b becomes a full-fledged primary and worker1a its secondary. glow up complete. 💖
$ docker compose exec monitor pg_autoctl show state
Name | Node | Host:Port | TLI: LSN | Connection | Reported State | Assigned State
---------+-------+---------------+----------------+--------------+---------------------+--------------------
coord0a | 0/3 | coord0a:5432 | 1: 0/3178B20 | read-write | primary | primary
coord0b | 0/4 | coord0b:5432 | 1: 0/3178B20 | read-only | secondary | secondary
worker1a | 1/1 | worker1a:5432 | 2: 0/504C400 | read-only | secondary | secondary
worker1b | 1/2 | worker1b:5432 | 2: 0/504C400 | read-write | primary | primary
worker2b | 2/7 | worker2b:5432 | 2: 0/50FF048 | read-only | secondary | secondary
worker2a | 2/8 | worker2a:5432 | 2: 0/50FF048 | read-write | primary | primary
worker3a | 3/5 | worker3a:5432 | 1: 0/31CD8C0 | read-write | primary | primary
worker3b | 3/6 | worker3b:5432 | 1: 0/31CD8C0 | read-only | secondary | secondaryLOOK AT THAT TABLE. 🦋 Every node, its group, its Host:Port, its TLI: LSN, its Connection, and both its Reported State and Assigned State — all lined up and gorgeous. 💅 This is the whole squad, bestie, and the vibes are immaculate.
💥 Handle Coordinator Failure 💥
Because our application connection string includes both coordinator hosts with the option target_session_attrs=read-write, the database client will connect to whichever of these servers supports both reads and writes. ✨ That's the whole point of the trick, no cap — the client just finds the one that does both and goes there.
However if we use the same trick with the pg_control file to crash our primary coordinator, we can watch how the monitor promotes the secondary. 👑
$ docker compose exec coord0a rm /tmp/pgaf/global/pg_control
$ docker compose exec coord0a /usr/lib/postgresql/14/bin/pg_ctl stop -D /tmp/pgaf🫶 Real talk: we are straight up deleting the pg_control file and then stopping the server. That's the deliberate crash recipe. Do not do this to anything you love.
After some time, coordinator A's keeper heals it, and the cluster converges in this state:
$ docker compose exec monitor pg_autoctl show state
Name | Node | Host:Port | TLI: LSN | Connection | Reported State | Assigned State
---------+-------+---------------+----------------+--------------+---------------------+--------------------
coord0a | 0/3 | coord0a:5432 | 2: 0/50000D8 | read-only | secondary | secondary
coord0b | 0/4 | coord0b:5432 | 2: 0/50000D8 | read-write | primary | primary
worker1a | 1/1 | worker1a:5432 | 2: 0/504C520 | read-only | secondary | secondary
worker1b | 1/2 | worker1b:5432 | 2: 0/504C520 | read-write | primary | primary
worker2b | 2/7 | worker2b:5432 | 2: 0/50FF130 | read-only | secondary | secondary
worker2a | 2/8 | worker2a:5432 | 2: 0/50FF130 | read-write | primary | primary
worker3a | 3/5 | worker3a:5432 | 1: 0/31CD8C0 | read-write | primary | primary
worker3b | 3/6 | worker3b:5432 | 1: 0/31CD8C0 | read-only | secondary | secondaryNot me spotting that coord0a is now read-only and secondary, while coord0b stepped up to read-write and primary. 👑 The glow-up is real. The monitor handled it, the keeper healed the poor crashed node, and everything converged like nothing ever happened.
We can check that the data is still available through the new coordinator node too:
$ docker compose exec app psql -c 'select count(*) from companies;'
count
-------
75SEVENTY-FIVE. 🎀 The data said "I'm still here bestie" and it was not lying.
🧹 Cleanup 🧹
To dispose of the entire tutorial environment, just use the following command:
$ docker compose downPoof. 💨 Gone. Clean slate, main character energy.
🌈 Next steps 🌈
As mentioned in the first section of this tutorial, the way we use docker compose here is not meant to be production ready. It's useful to understand and play with a distributed system such as Citus though, and makes it simple to introduce faults and see how the pg_auto_failover High Availability reacts to those faults.
One obvious missing element to better test the system is the lack of persistent volumes in our docker compose based test rig. It is possible to create external volumes and use them for each node in the docker compose definition. This allows restarting nodes over the same data set.
See the command pg_autoctl do tmux compose session for more details about how to run a docker compose test environment with docker compose, including external volumes for each node.
Now is a good time to go read Citus Documentation too, so that you know how to use this cluster you just created! 📚 Go read it, bestie, you earned it.
📖 MANUAL PAGES 📖
The pg_autoctl tool hosts many commands and sub-commands. Each of them have their own manual page. (Yes, they all have their own little moment. Deserved.)
🎀 pg_autoctl 🎀
pg_autoctl - control a pg_auto_failover node
✨ Synopsis ✨
pg_autoctl provides the following commands:
pg_autoctl
+ create Create a pg_auto_failover node, or formation
+ drop Drop a pg_auto_failover node, or formation
+ config Manages the pg_autoctl configuration
+ show Show pg_auto_failover information
+ enable Enable a feature on a formation
+ disable Disable a feature on a formation
+ get Get a pg_auto_failover node, or formation setting
+ set Set a pg_auto_failover node, or formation setting
+ perform Perform an action orchestrated by the monitor
activate Activate a Citus worker from the Citus coordinator
run Run the pg_autoctl service (monitor or keeper)
stop signal the pg_autoctl service for it to stop
reload signal the pg_autoctl for it to reload its configuration
status Display the current status of the pg_autoctl service
help print help message
version print pg_autoctl version
pg_autoctl create
monitor Initialize a pg_auto_failover monitor node
postgres Initialize a pg_auto_failover standalone postgres node
coordinator Initialize a pg_auto_failover citus coordinator node
worker Initialize a pg_auto_failover citus worker node
formation Create a new formation on the pg_auto_failover monitor
pg_autoctl drop
monitor Drop the pg_auto_failover monitor
node Drop a node from the pg_auto_failover monitor
formation Drop a formation on the pg_auto_failover monitor
pg_autoctl config
check Check pg_autoctl configuration
get Get the value of a given pg_autoctl configuration variable
set Set the value of a given pg_autoctl configuration variable
pg_autoctl show
uri Show the postgres uri to use to connect to pg_auto_failover nodes
events Prints monitor's state of nodes in a given formation and group
state Prints monitor's state of nodes in a given formation and group
settings Print replication settings for a formation from the monitor
standby-names Prints synchronous_standby_names for a given group
file List pg_autoctl internal files (config, state, pid)
systemd Print systemd service file for this node
pg_autoctl enable
secondary Enable secondary nodes on a formation
maintenance Enable Postgres maintenance mode on this node
ssl Enable SSL configuration on this node
monitor Enable a monitor for this node to be orchestrated from
pg_autoctl disable
secondary Disable secondary nodes on a formation
maintenance Disable Postgres maintenance mode on this node
ssl Disable SSL configuration on this node
monitor Disable the monitor for this node
pg_autoctl get
+ node get a node property from the pg_auto_failover monitor
+ formation get a formation property from the pg_auto_failover monitor
pg_autoctl get node
replication-quorum get replication-quorum property from the monitor
candidate-priority get candidate property from the monitor
pg_autoctl get formation
settings get replication settings for a formation from the monitor
number-sync-standbys get number_sync_standbys for a formation from the monitor
pg_autoctl set
+ node set a node property on the monitor
+ formation set a formation property on the monitor
pg_autoctl set node
metadata set metadata on the monitor
replication-quorum set replication-quorum property on the monitor
candidate-priority set candidate property on the monitor
pg_autoctl set formation
number-sync-standbys set number-sync-standbys for a formation on the monitor
pg_autoctl perform
failover Perform a failover for given formation and group
switchover Perform a switchover for given formation and group
promotion Perform a failover that promotes a target nodeThat's the whole family tree right there. 🌳 Every single one of these has its own manual page, iykyk.
💜 Description 💜
The pg_autoctl tool is the client tool provided by pg_auto_failover to create and manage Postgres nodes and the pg_auto_failover monitor node. The command is built with many sub-commands that each have their own manual page.
🫶 Help 🫶
To get the full recursive list of supported commands, use:
pg_autoctl helpWhen you're lost, just ask for help. It's giving self-care. 💅
⭐ Version ⭐
To grab the version of pg_autoctl that you're using, use:
pg_autoctl --version
pg_autoctl versionA typical output would be:
pg_autoctl version 1.4.2
pg_autoctl extension version 1.4
compiled with PostgreSQL 12.3 on x86_64-apple-darwin16.7.0, compiled by Apple LLVM version 8.1.0 (clang-802.0.42), 64-bit
compatible with Postgres 10, 11, 12, and 13The version is also available as a JSON document when using the --json option:
pg_autoctl --version --json
pg_autoctl version --jsonA typical JSON output would be:
{
"pg_autoctl": "1.4.2",
"pgautofailover": "1.4",
"pg_major": "12",
"pg_version": "12.3",
"pg_version_str": "PostgreSQL 12.3 on x86_64-apple-darwin16.7.0, compiled by Apple LLVM version 8.1.0 (clang-802.0.42), 64-bit",
"pg_version_num": 120003
}This is for version 1.4.2 of pg_auto_failover. This particular version of the pg_autoctl client tool has been compiled using libpq for PostgreSQL 12.3 and is compatible with Postgres 10, 11, 12, and 13. ✨
🌸 Environment 🌸
PG_AUTOCTL_DEBUG— When this environment variable is set (to anything) thenpg_autoctlallows more commands. Use with care, this opens abilities to destroy your production clusters.
🚨 BESTIE. HEAR ME. 🚨 "Use with care, this opens abilities to destroy your production clusters" is not a vibe, it's a threat. Set this and you unlock the extra commands. Only do it if you know exactly what you're doing. Not me risking the whole prod cluster for a little debugging moment. 💀
🦋 pg_autoctl create 🦋
pg_autoctl create - Create a pg_auto_failover node, or formation
👑 pg_autoctl create monitor 👑
pg_autoctl create monitor - Initialize a pg_auto_failover monitor node
✨ Synopsis ✨
This command initializes a PostgreSQL cluster and installs the pgautofailover extension so that it's possible to use the new instance to monitor PostgreSQL services:
usage: pg_autoctl create monitor [ --pgdata --pgport --pgctl --hostname ]And that's where this part leaves you — with the monitor-creation command and its args all lined up, waiting for the rest of the story. 🎀
--pgctl path to pg_ctl
--pgdata path to data directory
--pgport PostgreSQL's port number
--hostname hostname by which postgres is reachable
--auth authentication method for connections from data nodes
--skip-pg-hba skip editing pg_hba.conf rules
--run create node then run pg_autoctl service
--ssl-self-signed setup network encryption using self signed certificates (does NOT protect against MITM)
--ssl-mode use that sslmode in connection strings
--ssl-ca-file set the Postgres ssl_ca_file to that file path
--ssl-crl-file set the Postgres ssl_crl_file to that file path
--no-ssl don't enable network encryption (NOT recommended, prefer --ssl-self-signed)
--server-key set the Postgres ssl_key_file to that file path
--server-cert set the Postgres ssl_cert_file to that file pathDescription
The pg_autoctl tool is literally the client tool that pg_auto_failover hands you — you use it to create and manage your Postgres nodes and the pg_auto_failover monitor node. It's built out of a whole bunch of sub-commands, and every single one of them has its own manual page, so if you need the deep lore on a specific one, go find its page. 👑
Options
Okay bestie, here's the full menu of options you can pass to pg_autoctl create monitor. Every one of them, no skipping. 🌸
--pgctl— The path to thepg_ctltool you wanna use, for the version of PostgreSQL you're running.
By default it just grabs the pg_ctl it finds in your PATH, but only when there's a single entry for pg_ctl in PATH. Double-check your setup with which -a pg_ctl.
On an RPM-based distro like RHEL or CentOS, that path is usually /usr/pgsql-13/bin/pg_ctl for Postgres 13.
On a Debian-based distro like debian or ubuntu, it's usually /usr/lib/postgresql/13/bin/pg_ctl for Postgres 13. Those distros also ship the package postgresql-common, which gives you /usr/bin/pg_config. And pg_autoctl can use that tool automatically to discover the default version of Postgres on your setup. ✨
--pgdata— Where to initialize a Postgres database cluster, using eitherpg_ctl initdborpg_basebackup. Defaults to the environment variablePGDATA.--pgport— The Postgres port to use. Defaults to5432.--hostname— The hostname or IP address (both v4 and v6 are supported 💅) that every other node should use to connect to this node.
If you don't provide one, a default value gets computed by running this exact algorithm:
- We get this machine's "public IP" by opening a connection to the
8.8.8.8:53public service. Then we grab the TCP/IP client address that was used to make that connection. - We then do a reverse DNS lookup on the IP address found in the previous step to fetch a hostname for our local machine.
- If the reverse DNS lookup is successful, then
pg_autoctldoes a forward DNS lookup of that hostname.
When the forward DNS lookup response in step 3. is an IP address found in one of our local network interfaces, then pg_autoctl uses the hostname found in step 2. as the default --hostname. Otherwise it uses the IP address found in step 1.
You can also just use the --hostname command line option to bypass the whole DNS-lookup-based process and force the local node name to a fixed value. 🦋
--auth— The authentication methodpg_autoctluses when editing the Postgres HBA file to open connections to other nodes. No default value, you must provide one. The value--trustis only a good choice for testing and evaluation ofpg_auto_failover— see Security settings for pg_auto_failover for more info. 🔥--skip-pg-hba— When you use this,pg_autoctlrefrains from any editing of the Postgres HBA file. But heads up: editing the HBA file is still needed so other nodes can connect using either read privileges or replication streaming privileges. 👀
When --skip-pg-hba is used, pg_autoctl still outputs the HBA entries it needs in the logs — it only skips editing the HBA file.
--run— Immediately run thepg_autoctlservice after having created this node.--ssl-self-signed— Generate SSL self-signed certificates to provide network encryption. This does not protect against man-in-the-middle kinds of attacks. See Security settings for pg_auto_failover for more about our SSL settings.--ssl-mode— The SSL Mode used bypg_autoctlwhen connecting to other nodes, including when connecting for streaming replication.--ssl-ca-file— Set the Postgresssl_ca_fileto that file path.--ssl-crl-file— Set the Postgresssl_crl_fileto that file path.--no-ssl— Don't enable network encryption. Not recommended, prefer--ssl-self-signed. 💀--server-key— Set the Postgresssl_key_fileto that file path.--server-cert— Set the Postgresssl_cert_fileto that file path.
Environment
Env vars are the vibes your shell passes down, and these are the ones that matter here. 🧚
PGDATA— Postgres directory location. Can be used instead of the--pgdataoption.PG_CONFIG— Can be set to the absolute path to thepg_configPostgres tool. This is mostly used in the context of building extensions, though it can be a useful way to select a Postgres version when several are installed on the same system.PATH— Used the usual way, mostly. Some entries that are searched in thePATHby thepg_autoctlcommand are expected to be found only once, to avoid mistakes with Postgres major versions.PGHOST,PGPORT,PGDATABASE,PGUSER,PGCONNECT_TIMEOUT, ... — See the Postgres docs about Environment Variables for details.TMPDIR— Thepgcopydbcommand creates all its work files and directories in${TMPDIR}/pgcopydb, and defaults to/tmp/pgcopydb.XDG_CONFIG_HOME— Thepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification.XDG_DATA_HOME— Thepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification.
🌈 pg_autoctl create postgres 🌈
pg_autoctl create postgres — Initialize a pg_auto_failover postgres node
Synopsis
The command pg_autoctl create postgres initializes a standalone Postgres node into a pg_auto_failover monitor. The monitor is then handling auto-failover for this Postgres node — but only as soon as a secondary has been registered too, and is known to be healthy. 👑
usage: pg_autoctl create postgres
--pgctl path to pg_ctl
--pgdata path to data directory
--pghost PostgreSQL's hostname
--pgport PostgreSQL's port number
--listen PostgreSQL's listen_addresses
--username PostgreSQL's username
--dbname PostgreSQL's database name
--name pg_auto_failover node name
--hostname hostname used to connect from the other nodes
--formation pg_auto_failover formation
--monitor pg_auto_failover Monitor Postgres URL
--auth authentication method for connections from monitor
--skip-pg-hba skip editing pg_hba.conf rules
--pg-hba-lan edit pg_hba.conf rules for --dbname in detected LAN
--ssl-self-signed setup network encryption using self signed certificates (does NOT protect against MITM)
--ssl-mode use that sslmode in connection strings
--ssl-ca-file set the Postgres ssl_ca_file to that file path
--ssl-crl-file set the Postgres ssl_crl_file to that file path
--no-ssl don't enable network encryption (NOT recommended, prefer --ssl-self-signed)
--server-key set the Postgres ssl_key_file to that file path
--server-cert set the Postgres ssl_cert_file to that file path
--candidate-priority priority of the node to be promoted to become primary
--replication-quorum true if node participates in write quorum
--maximum-backup-rate maximum transfer rate of data transferred from the server during initial syncDescription
Three different modes of initialization are supported by this command, corresponding to as many implementation strategies. Four of them are numbered below, but the intro said three, so like… read them all anyway, bestie. 😭 Not me pointing that out. Anyway:
- Initialize a primary node from scratch 🎀
This happens when --pgdata (or the environment variable PGDATA) points to a non-existing or empty directory. Then the given --hostname is registered to the pg_auto_failover --monitor as a member of the --formation.
The monitor answers to the registration call with a state to assign to the new member of the group, either SINGLE or WAIT_STANDBY. When the assigned state is SINGLE, then pg_autoctl create postgres proceeds to initialize a new PostgreSQL instance.
- Initialize an already existing primary server ✨
This happens when --pgdata (or the environment variable PGDATA) points to an already existing directory that belongs to a PostgreSQL instance. The standard PostgreSQL tool pg_controldata is used to recognize whether the directory belongs to a PostgreSQL instance.
In that case, the given --hostname is registered to the monitor in the tentative SINGLE state. When the given --formation and --group is currently empty, then the monitor accepts the registration and the pg_autoctl create prepares the already existing primary server for pg_auto_failover.
- Initialize a secondary node from scratch 🦋
This happens when --pgdata (or the environment variable PGDATA) points to a non-existing or empty directory, and when the monitor registration call assigns the state WAIT_STANDBY in step 1.
In that case, the pg_autoctl create command steps through the initial states of registering a secondary server, which includes preparing the primary server PostgreSQL HBA rules and creating a replication slot.
When the command ends successfully, a PostgreSQL secondary server has been created with pg_basebackup and is now started, catching-up to the primary server. Slay. 💅
- Initialize a secondary node from an existing data directory 🌸
When the data directory pointed to by the option --pgdata or the environment variable PGDATA already exists, then pg_auto_failover verifies that the system identifier matches the one of the other nodes already existing in the same group.
The system identifier? You can grab it with pg_controldata, bestie. And here's the tea: every node in a physical replication setup has to share the same system identifier, so in pg_auto_failover all the nodes in one group are stuck with that same constraint too. No exceptions. 💅
So when the system identifier matches one that's already registered by other nodes in the same group, the node gets set up as a standby, and Postgres boots up with the primary conninfo pointed straight at the current primary. Slay. ✨
The --auth option lets you set the authentication method used when the monitor node makes a connection to a data node as the pgautofailover_monitor user. Just like with the pg_autoctl create monitor command, you can totally use --auth trust while you're just playing around with pg_auto_failover at first, and then consider something production grade later. Also — big brain move — consider using --skip-pg-hba if you already have your own provisioning tools with a security compliance process. 👑
Peep Security settings for pg_auto_failover for the notes on .pgpass, okay? 🫶
✨ Options ✨
Here's everything you can throw at pg_autoctl create postgres, bestie:
--pgctl— Path to thepg_ctltool you wanna use for the version of PostgreSQL you're going for.
It defaults to the pg_ctl found in the PATH when there's only a single entry for pg_ctl in the PATH. Check your setup with which -a pg_ctl.
On an RPM based distribution like RHEL or CentOS, the path would usually be /usr/pgsql-13/bin/pg_ctl for Postgres 13.
On a debian based distribution like debian or ubuntu, the path would usually be /usr/lib/postgresql/13/bin/pg_ctl for Postgres 13. Those distros also use the package postgresql-common, which provides /usr/bin/pg_config. That tool can be automatically used by pg_autoctl to discover the default version of Postgres to use on your setup.
--pgdata— Location where you wanna initialize a Postgres database cluster, using eitherpg_ctl initdborpg_basebackup. Defaults to the environment variablePGDATA.--pghost— Hostname to use when connecting to the local Postgres instance from thepg_autoctlprocess. By default this field is left blank in the connection string, which lets you use Unix Domain Sockets with the default path compiled into your libpq version — usually provided by the Operating System. That would be/var/run/postgresqlwhen using debian or ubuntu.--pgport— Postgres port to use, defaults to5432.--listen— PostgreSQL'slisten_addressesto set up. Right now only one address is supported in this command line option, jsyk.--username— PostgreSQL's username to use when connecting to the local Postgres instance to manage it.--dbname— PostgreSQL's database name to use in your application. Defaults to being the same as the--username, or topostgreswhen none of those options are used.--name— Node name used on the monitor to refer to this node. The hostname is a technical information situation, and given Postgres requirements on the HBA setup and DNS resolution (both forward and reverse lookups), IP addresses are often used for the hostname.
The --name option lets you use a user-friendly name for your Postgres nodes instead. Cute, right?
--hostname— Hostname or IP address (both v4 and v6 are supported) to use from any other node to connect to this node.
When it's not provided, a default value gets computed by running this algorithm:
- We get this machine's "public IP" by opening a connection to the given monitor hostname or IP address. Then we get the TCP/IP client address that was used to make that connection.
- We then do a reverse DNS lookup on the IP address found in the previous step to fetch a hostname for our local machine.
- If the reverse DNS lookup is successful, then
pg_autoctldoes a forward DNS lookup of that hostname.
When the forward DNS lookup response in step 3 is an IP address found in one of our local network interfaces, then pg_autoctl uses the hostname found in step 2 as the default --hostname. Otherwise it uses the IP address found in step 1.
You may use the --hostname command line option to bypass the whole DNS lookup based process and force the local node name to a fixed value. Main character energy, honestly. 💅
--formation— Formation to register the node into on the monitor. Defaults to thedefaultformation, which is automatically created in the monitor in thepg_autoctl create monitorcommand.--monitor— Postgres URI used to connect to the monitor. Must use theautoctl_nodeusername and target thepg_auto_failoverdatabase name. It's possible to show the Postgres URI from the monitor node using the commandpg_autoctl show uri.--auth— Authentication method used bypg_autoctlwhen editing the Postgres HBA file to open connections to other nodes. No default value, must be provided by the user. The value--trustis only a good choice for testing and evaluation ofpg_auto_failover— see Security settings for pg_auto_failover for more information. 🚨--skip-pg-hba— When this option is used,pg_autoctlrefrains from any editing of the Postgres HBA file. Please note that editing the HBA file is still needed so that other nodes can connect using either read privileges or replication streaming privileges.
When --skip-pg-hba is used, pg_autoctl still outputs the HBA entries it needs in the logs — it only skips editing the HBA file.
--pg-hba-lan— When this option is used,pg_autoctldetermines the local IP address used to connect to the monitor, retrieves its netmask, and uses that to compute your local area network CIDR. This CIDR is then opened for connections in the Postgres HBA rules.
For instance, when the monitor resolves to 192.168.0.1 and your local Postgres node uses an interface with IP address 192.168.0.2/255.255.255.0 to connect to the monitor, then the LAN CIDR is computed to be 192.168.0.0/24.
--candidate-priority— Sets this node's replication setting for candidate priority to the given value (between0and100) at node registration on the monitor. Defaults to50.--replication-quorum— Sets this node's replication setting for replication quorum to the given value (eithertrueorfalse) at node registration on the monitor. Defaults totrue, which enables synchronous replication.--maximum-backup-rate— Sets the maximum transfer rate of data transferred from the server during initial sync. This is used bypg_basebackup. Defaults to100M.--run— Immediately run thepg_autoctlservice after having created this node.--ssl-self-signed— Generate SSL self-signed certificates to provide network encryption. This does not protect against man-in-the-middle kinds of attacks. See Security settings for pg_auto_failover for more about our SSL settings. ⚠️--ssl-mode— SSL Mode used bypg_autoctlwhen connecting to other nodes, including when connecting for streaming replication.--ssl-ca-file— Set the Postgresssl_ca_fileto that file path.--ssl-crl-file— Set the Postgresssl_crl_fileto that file path.--no-ssl— Don't enable network encryption. This is not recommended, prefer--ssl-self-signed.--server-key— Set the Postgresssl_key_fileto that file path.--server-cert— Set the Postgresssl_cert_fileto that file path.
🌱 Environment 🌱
Env vars are the vibe too, bestie. Here's the lineup:
PGDATA— Postgres directory location. Can be used instead of the--pgdataoption.PG_AUTOCTL_MONITOR— Postgres URI to connect to the monitor node, can be used instead of the--monitoroption.PG_AUTOCTL_NODE_NAME— Node name to register to the monitor, can be used instead of the--nameoption.PG_AUTOCTL_REPLICATION_QUORUM— Can be used instead of the--replication-quorumoption.PG_AUTOCTL_CANDIDATE_PRIORITY— Can be used instead of the--candidate-priorityoption.PG_CONFIG— Can be set to the absolute path to thepg_configPostgres tool. This is mostly used in the context of building extensions, though it can be a useful way to select a Postgres version when several are installed on the same system.PATH— Used the usual way mostly. Some entries that are searched in thePATHby thepg_autoctlcommand are expected to be found only once, to avoid mistakes with Postgres major versions.PGHOST,PGPORT,PGDATABASE,PGUSER,PGCONNECT_TIMEOUT, ...` — See the Postgres docs about Environment Variables for details.TMPDIR— Thepgcopydbcommand creates all its work files and directories in${TMPDIR}/pgcopydb, and defaults to/tmp/pgcopydb.XDG_CONFIG_HOME— Thepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification.XDG_DATA_HOME— Thepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification.
👑 pg_autoctl create coordinator 👑
pg_autoctl create coordinator — Initialize a pg_auto_failover coordinator node
✨ Synopsis ✨
The command pg_autoctl create coordinator initializes a pg_auto_failover Coordinator node for a Citus formation. The coordinator is special in a Citus formation: that's where the client application connects to either manage the formation and the sharding of the tables, or for its normal SQL traffic.
The coordinator also has to register every worker in the formation. Periodt. 💖
usage: pg_autoctl create coordinator --pgctl path to pg_ctl
--pgdata path to data directory
--pghost PostgreSQL's hostname
--pgport PostgreSQL's port number
--hostname hostname by which postgres is reachable
--listen PostgreSQL's listen_addresses
--username PostgreSQL's username
--dbname PostgreSQL's database name
--name pg_auto_failover node name
--formation pg_auto_failover formation
--monitor pg_auto_failover Monitor Postgres URL
--auth authentication method for connections from monitor
--skip-pg-hba skip editing pg_hba.conf rules
--citus-secondary when used, this worker node is a citus secondary
--citus-cluster name of the Citus Cluster for read-replicas
--ssl-self-signed setup network encryption using self signed certificates (does NOT protect against MITM)
--ssl-mode use that sslmode in connection strings
--ssl-ca-file set the Postgres ssl_ca_file to that file path
--ssl-crl-file set the Postgres ssl_crl_file to that file path
--no-ssl don't enable network encryption (NOT recommended, prefer --ssl-self-signed)
--server-key set the Postgres ssl_key_file to that file path
--server-cert set the Postgres ssl_cert_file to that file pathDescription 🌸
Okay bestie, this command is literally doing the exact same thing as pg_autoctl create postgres, no cap — but it tacks on these extra steps, and they're kind of the whole point:
- adds
shared_preload_libraries = citusto the local PostgreSQL instance configuration. - enables the whole local area network to connect to the coordinator, by adding an entry for e.g.
192.168.1.0/24in the PostgreSQL HBA configuration. - creates the extension
citusin the target database.
IMPORTANT: 💀 The default--dbnameis the same as the current system user name, which in many case is going to bepostgres. Please make sure to use the--dbnameoption with the actual database that you're going to use with your application. Citus does not support multiple databases, you have to use the database where Citus is created. When using Citus, that is essential to the well behaving of worker failover.
Options 💅
Go read the manual page for pg_autoctl create postgres for the common options, okay? This bit right here only lists what's special to pg_autoctl create coordinator:
--citus-secondary— Use this option to create a coordinator dedicated to a Citus Secondary cluster. See Citus Secondaries and read-replica for more information.--citus-cluster— Use this option to name the Citus Secondary cluster that this coordinator node belongs to. Use the same cluster name again for the worker nodes that are part of this cluster. See Citus Secondaries and read-replica for more information.
🦋 pg_autoctl create worker
Initialize a pg_auto_failover worker node
Synopsis ✨
The command pg_autoctl create worker initializes a pg_auto_failover Worker node for a Citus formation. The worker is special in a Citus formation: that's where the client application connects to either to manage the formation and the sharding of the tables, or for its normal SQL traffic.
The worker also has to register every worker in the formation.
usage: pg_autoctl create worker
--pgctl path to pg_ctl
--pgdata path to data director
--pghost PostgreSQL's hostname
--pgport PostgreSQL's port number
--hostname hostname by which postgres is reachable
--listen PostgreSQL's listen_addresses
--proxyport Proxy's port number
--username PostgreSQL's username
--dbname PostgreSQL's database name
--name pg_auto_failover node name
--formation pg_auto_failover formation
--group pg_auto_failover group Id
--monitor pg_auto_failover Monitor Postgres URL
--auth authentication method for connections from monitor
--skip-pg-hba skip editing pg_hba.conf rules
--citus-secondary when used, this worker node is a citus secondary
--citus-cluster name of the Citus Cluster for read-replicas
--ssl-self-signed setup network encryption using self signed certificates (does NOT protect against MITM)
--ssl-mode use that sslmode in connection strings
--ssl-ca-file set the Postgres ssl_ca_file to that file path
--ssl-crl-file set the Postgres ssl_crl_file to that file path
--no-ssl don't enable network encryption (NOT recommended, prefer --ssl-self-signed)
--server-key set the Postgres ssl_key_file to that file path
--server-cert set the Postgres ssl_cert_file to that file pathDescription 🌈
Same energy as pg_autoctl create postgres, bestie — but with these extra steps layered on top:
- adds
shared_preload_libraries = citusto the local PostgreSQL instance configuration. - creates the extension
citusin the target database. - gets the coordinator node hostname from the pg_auto_failover monitor.
This operation is retried when it fails, as the coordinator might appear later than some of the workers when the whole formation is initialized at once, in parallel, on multiple nodes.
- adds node to the coordinator
This is done in two steps. First, we call the SQL function master_add_inactive_node on the coordinator, and second, we call the SQL function master_activate_node.
This way allows for easier diagnostics in case things go wrong. In the first step, the network and authentication setup needs to allow for nodes to connect to each other. In the second step, the Citus reference tables are distributed to the new node, and this operation has its own set of failure cases to handle.
IMPORTANT: 😭 The default--dbnameis the same as the current system user name, which in many case is going to bepostgres. Please make sure to use the--dbnameoption with the actual database that you're going to use with your application. Citus does not support multiple databases, you have to use the database where Citus is created. When using Citus, that is essential to the well behaving of worker failover.
Options 🎀
Again — see the manual page for pg_autoctl create postgres for the common options. What follows is only what's specific to pg_autoctl create worker:
--proxyport— The--proxyportoption allows pg_auto_failover to register the proxy port in thepg_dist_poolinfoentry for the worker node in its Coordinator, rather than the--pgportentry as would usually be done.--citus-secondary— Use this option to create a worker dedicated to a Citus Secondary cluster. See Citus Secondaries and read-replica for more information.--citus-cluster— Use this option to name the Citus Secondary cluster that this worker node belongs to. Use the same cluster name again for the worker nodes that are part of this cluster. See Citus Secondaries and read-replica for more information.
💜 pg_autoctl create formation
Create a new formation on the pg_auto_failover monitor
Synopsis 🌟
This command registers a new formation on the monitor, with the specified kind:
usage: pg_autoctl create formation [ --pgdata --monitor --formation --kind --dbname --with-secondary --without-secondary ]
--pgdata path to data directory
--monitor pg_auto_failover Monitor Postgres URL
--formation name of the formation to create
--kind formation kind, either "pgsql" or "citus"
--dbname name for postgres database to use in this formation
--enable-secondary create a formation that has multiple nodes that can be
used for fail over when others have issues
--disable-secondary create a citus formation without nodes to fail over to
--number-sync-standbys minimum number of standbys to confirm writeDescription 🫶
A single pg_auto_failover monitor may manage any number of formations, each composed of at least one Postgres service group. This commands creates a new formation so that it is then possible to register Postgres nodes in the new formation.
Options 🍓
The following options are available to pg_autoctl create formation:
--pgdata— Location where to initialize a Postgres database cluster, using eitherpg_ctl initdborpg_basebackup. Defaults to the environment variablePGDATA.--monitor— Postgres URI used to connect to the monitor. Must use theautoctl_nodeusername and target thepg_auto_failoverdatabase name. It is possible to show the Postgres URI from the monitor node using the commandpg_autoctl show uri.--formation— Name of the formation to create.--kind— A pg_auto_failover formation could be of kindpgsqlor of kindcitus. At the moment citus formation kinds are not managed in the Open Source version of pg_auto_failover.--dbname— Name of the database to use in the formation, mostly useful to formation kinds citus where the Citus extension is only installed in a single target database.--enable-secondary— The formation to be created allows using standby nodes. Defaults totrue. Mostly useful for Citus formations.--disable-secondary— See--enable-secondaryabove.--number-sync-standby— Postgres streaming replication usessynchronous_standby_namesto setup how many standby nodes should have received a copy of the transaction data. When using pg_auto_failover this setup is handled at the formation level.
Defaults to zero when creating the first two Postgres nodes in a formation in the same group. When set to zero pg_auto_failover uses synchronous replication only when a standby node is available: the idea is to allow failover, this setting does not allow proper HA for Postgres.
When adding a third node that participates in the quorum (one primary, two secondaries), the setting is automatically changed from zero to one.
Environment 🌸
PGDATA— Postgres directory location. Can be used instead of the--pgdataoption.PG_AUTOCTL_MONITOR— Postgres URI to connect to the monitor node, can be used instead of the--monitoroption.XDG_CONFIG_HOME— bestie, this is wherepg_autoctlparks its config files, in the standard spot. See the XDG Base Directory Specification for the deets. ✨XDG_DATA_HOME— and this one is wherepg_autoctlstashes its internal state files, again in the standard place. It defaults to~/.local/share. See the XDG Base Directory Specification. 💾
🗑️ pg_autoctl drop
pg_autoctl drop — Drop a pg_auto_failover node, or formation
💅 pg_autoctl drop monitor
pg_autoctl drop monitor — Drop the pg_auto_failover monitor
✨ Synopsis
This command allows to review all the replication settings of a given formation (defaults to 'default' as usual):
usage: pg_autoctl drop monitor [ --pgdata --destroy ]
--pgdata path to data directory
--destroy also destroy Postgres database🎀 Options
--pgdata— Location of the Postgres node being managed locally. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl.--destroy— By default thepg_autoctl drop monitorcommands does not remove the Postgres database for the monitor. When using--destroy, the Postgres installation is also deleted.
🌸 Environment
PGDATA— Postgres directory location. Can be used instead of the--pgdataoption.XDG_CONFIG_HOME— Thepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification.XDG_DATA_HOME— Thepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification.
🦋 pg_autoctl drop node
pg_autoctl drop node — Drop a node from the pg_auto_failover monitor
✨ Synopsis
This command drops a Postgres node from the pg_auto_failover monitor:
usage: pg_autoctl drop node [ [ [ --pgdata ] [ --destroy ] ] | [ --monitor [ [ --hostname --pgport ] | [ --formation --name ] ] ] ]
--pgdata path to data directory
--monitor pg_auto_failover Monitor Postgres URL
--formation pg_auto_failover formation
--name drop the node with the given node name
--hostname drop the node with given hostname and pgport
--pgport drop the node with given hostname and pgport
--destroy also destroy Postgres database
--force force dropping the node from the monitor
--wait how many seconds to wait, default to 60💖 Description
Two modes of operations are implemented in the pg_autoctl drop node command.
When removing a node that still exists, it is possible to use pg_autoctl drop node --destroy to remove the node both from the monitor and also delete the local Postgres instance entirely.
When removing a node that doesn't exist physically anymore, or when the VM that used to host the node has been lost entirely, use either the pair of options --hostname and --pgport or the pair of options --formation and --name to match the node registration record on the monitor database, and get it removed from the known list of nodes on the monitor.
Then option --force can be used when the target node to remove does not exist anymore. When a node has been lost entirely, it's not going to be able to finish the procedure itself, and it is then possible to instruct the monitor of the situation.
🎀 Options
--pgdata— Location of the Postgres node being managed locally. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl.--monitor— Postgres URI used to connect to the monitor. Must use theautoctl_nodeusername and target thepg_auto_failoverdatabase name. It is possible to show the Postgres URI from the monitor node using the commandpg_autoctl show uri.--hostname— Hostname of the Postgres node to remove from the monitor. Use either--nameor--hostname --pgport, but not both.--pgport— Port of the Postgres node to remove from the monitor. Use either--nameor--hostname --pgport, but not both.--name— Name of the node to remove from the monitor. Use either--nameor--hostname --pgport, but not both.--destroy— By default thepg_autoctl drop monitorcommands does not remove the Postgres database for the monitor. When using--destroy, the Postgres installation is also deleted.--force— By default a node is expected to reach the assigned stateDROPPEDwhen it is removed from the monitor, and has the opportunity to implement clean-up actions. When the target node to remove is not available anymore, it is possible to use the option--forceto immediately remove the node from the monitor.--wait— How many seconds to wait for the node to be dropped entirely. The command stops when the target node is not to be found on the monitor anymore, or when the timeout has elapsed, whichever comes first. The value0(zero) disables the timeout and disables waiting entirely, making the command async.
🌸 Environment
PGDATA— Postgres directory location. Can be used instead of the--pgdataoption.PG_AUTOCTL_MONITOR— Postgres URI to connect to the monitor node, can be used instead of the--monitoroption.PG_AUTOCTL_NODE_NAME— Node name to register to the monitor, can be used instead of the--nameoption.PG_AUTOCTL_REPLICATION_QUORUM— Can be used instead of the--replication-quorumoption.PG_AUTOCTL_CANDIDATE_PRIORITY— Can be used instead of the--candidate-priorityoption.PG_CONFIG— Can be set to the absolute path to thepg_configPostgres tool. This is mostly used in the context of building extensions, though it can be a useful way to select a Postgres version when several are installed on the same system.PATH— Used the usual way mostly. Some entries that are searched in thePATHby thepg_autoctlcommand are expected to be found only once, to avoid mistakes with Postgres major versions.PGHOST,PGPORT,PGDATABASE,PGUSER,PGCONNECT_TIMEOUT, ... — See the Postgres docs about Environment Variables for details.TMPDIR— Thepgcopydbcommand creates all its work files and directories in${TMPDIR}/pgcopydb, and defaults to/tmp/pgcopydb.XDG_CONFIG_HOME— Thepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification.XDG_DATA_HOME— Thepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification.
🔥 Examples
$ pg_autoctl drop node --destroy --pgdata ./node3
17:52:21 54201 INFO Reaching assigned state "secondary"
17:52:21 54201 INFO Removing node with name "node3" in formation "default" from the monitor
17:52:21 54201 WARN Postgres is not running and we are in state secondary
17:52:21 54201 WARN Failed to update the keeper's state from the local PostgreSQL instance, see above for details.
17:52:21 54201 INFO Calling node_active for node default/4/0 with current state: PostgreSQL is running is false, sync_state is "", latest WAL LSN is 0/0.
17:52:21 54201 INFO FSM transition to "dropped": This node is being dropped from the monitor
17:52:21 54201 INFO Transition complete: current state is now "dropped"
17:52:21 54201 INFO This node with id 4 in formation "default" and group 0 has been dropped from the monitor
17:52:21 54201 INFO Stopping PostgreSQL at "/Users/dim/dev/MS/pg_auto_failover/tmux/node3"
17:52:21 54201 INFO /Applications/Postgres.app/Contents/Versions/12/bin/pg_ctl --pgdata /Users/dim/dev/MS/pg_auto_failover/tmux/node3 --wait stop --mode fast
17:52:21 54201 INFO /Applications/Postgres.app/Contents/Versions/12/bin/pg_ctl status -D /Users/dim/dev/MS/pg_auto_failover/tmux/node3 [3]
17:52:21 54201 INFO pg_ctl: no server running
17:52:21 54201 INFO pg_ctl stop failed, but PostgreSQL is not running anyway
17:52:21 54201 INFO Removing "/Users/dim/dev/MS/pg_auto_failover/tmux/node3"
17:52:21 54201 INFO Removing "/Users/dim/dev/MS/pg_auto_failover/tmux/config/pg_autoctl/Users/dim/dev/MS/pg_auto_failover/tmux/node3/pg_autoctl.cfg"🧚 pg_autoctl drop formation
pg_autoctl drop formation — Drop a formation on the pg_auto_failover monitor
✨ Synopsis
This command drops an existing formation on the monitor:
usage: pg_autoctl drop formation [ --pgdata --formation ]
--pgdata path to data directory
--monitor pg_auto_failover Monitor Postgres URL
--formation name of the formation to drop🎀 Options
--pgdata— Location of the Postgres node being managed locally. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl.--monitor— Postgres URI used to connect to the monitor. Must use theautoctl_nodeusername and target thepg_auto_failoverdatabase name. It is possible to show the Postgres URI from the monitor node using the commandpg_autoctl show uri.--formation— Name of the formation to drop from the monitor.
🌸 Environment
PGDATA— Postgres directory location. Can be used instead of the--pgdataoption.PG_AUTOCTL_MONITOR— Postgres URI to connect to the monitor node, can be used instead of the--monitoroption.XDG_CONFIG_HOME— Thepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification.XDG_DATA_HOME— Thepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification.
⚙️ pg_autoctl config
pg_autoctl config — Manages the pg_autoctl configuration
👑 pg_autoctl config get
pg_autoctl config get — Get the value of a given pg_autoctl configuration variable
✨ Synopsis
This command prints a pg_autoctl configuration setting:
usage: pg_autoctl config get [ --pgdata ] [ --json ] [ section.option ]
--pgdata path to data directory🎀 Options
--pgdata— Location of the Postgres node being managed locally. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl.--json— Output JSON formatted data.
🌸 Environment 🌸
Okay bestie, these are the env vars that do stuff behind the scenes, and knowing them is lowkey a superpower:
PGDATA— where the Postgres directory lives. You can use this instead of the--pgdataoption. 💅PG_AUTOCTL_MONITOR— the Postgres URI for connecting to the monitor node. Works instead of the--monitoroption. ✨XDG_CONFIG_HOME— thepg_autoctlcommand keeps its config files in the standard place,XDG_CONFIG_HOME. See the XDG Base Directory Specification. 🎀XDG_DATA_HOME— thepg_autoctlcommand keeps its internal state files in the standard place,XDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification. 🦋
💖 Description 💖
When you pass the section.option argument, that's the name of a configuration option, periodt. The pg_autoctl config file is stored using the INI format — yes, classic, iconic.
When you give pg_autoctl config get no argument at all, it just dumps the entire config file at you. And if you're sitting there like "okay but where even is the config file", check pg_autoctl show file and use pg_autoctl show file --config. iykyk 🌈
✨ Examples ✨
No arguments? You get the whole file, bestie:
$ pg_autoctl config get --pgdata node1
[pg_autoctl]
role = keeper
monitor = postgres://autoctl_node@localhost:5500/pg_auto_failover?sslmode=prefer
formation = default
group = 0
name = node1
hostname = localhost
nodekind = standalone
[postgresql]
pgdata = /Users/dim/dev/MS/pg_auto_failover/tmux/node1
pg_ctl = /Applications/Postgres.app/Contents/Versions/12/bin/pg_ctl
dbname = demo
host = /tmp
port = 5501
proxyport = 0
listen_addresses = *
auth_method = trust
hba_level = app
[ssl]
active = 1
sslmode = require
cert_file = /Users/dim/dev/MS/pg_auto_failover/tmux/node1/server.crt
key_file = /Users/dim/dev/MS/pg_auto_failover/tmux/node1/server.key
[replication]
maximum_backup_rate = 100M
backup_directory = /Users/dim/dev/MS/pg_auto_failover/tmux/backup/node_1
[timeout]
network_partition_timeout = 20
prepare_promotion_catchup = 30
prepare_promotion_walreceiver = 5
postgresql_restart_failure_timeout = 20
postgresql_restart_failure_max_retries = 3And the move? Pipe the JSON output into the jq command line and filter the result down to just one section of the file. Chef's kiss 👑
$ pg_autoctl config get --pgdata node1 --json | jq .pg_autoctl
{
"role": "keeper",
"monitor": "postgres://autoctl_node@localhost:5500/pg_auto_failover?sslmode=prefer",
"formation": "default",
"group": 0,
"name": "node1",
"hostname": "localhost",
"nodekind": "standalone"
}Finally — because the glow-up never stops — you can list a single config element all by itself:
$ pg_autoctl config get --pgdata node1 ssl.sslmode --json
require👑 pg_autoctl config set 👑
pg_autoctl config set — Set the value of a given pg_autoctl configuration variable. Let's gooo 💅
⭐ Synopsis ⭐
This command prints a pg_autoctl configuration setting:
usage: pg_autoctl config set [ --pgdata ] [ --json ] section.option [ value ]
--pgdata path to data directory🎀 Options 🎀
--pgdata— Location of the Postgres node being managed locally. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl.--json— Output JSON formatted data.
🌸 Environment 🌸
PGDATA— Postgres directory location. Can be used instead of the--pgdataoption. ✨PG_AUTOCTL_MONITOR— Postgres URI to connect to the monitor node, can be used instead of the--monitoroption. 🦋XDG_CONFIG_HOME— Thepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification. 🎀XDG_DATA_HOME— Thepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification. 💜
💖 Description 💖
This command lets you set a pg_autoctl configuration setting to a shiny new value. Most settings can be changed, and most can be reloaded online — no drama, no downtime (well, mostly). ✨
Some of those commands can then be applied with a pg_autoctl reload command to an already running process. Slay.
🫶 Settings 🫶
pg_autoctl.role
This setting can not be changed. It can be either monitor or keeper, and the rest of the configuration file is read depending on this value. It's giving "the vibe decides everything downstream". 💀
pg_autoctl.monitor
URI of the pg_autoctl monitor Postgres service. Can be changed with a reload.
To register an existing node to a new monitor, use pg_autoctl disable monitor and then pg_autoctl enable monitor. Two-step process, but it works. 👑
pg_autoctl.formation
Formation to which this node has been registered. Changing this setting is not supported. 🙈
pg_autoctl.group
Group in which this node has been registered. Changing this setting is not supported. 🙈
pg_autoctl.name
Name of the node as known to the monitor and listed in pg_autoctl show state. Can be changed with a reload.
pg_autoctl.hostname
Hostname or IP address of the node, as known to the monitor. Can be changed with a reload.
pg_autoctl.nodekind
This setting can not be changed and depends on the command that has been used to create this pg_autoctl node. It is what it is. 💅
postgresql.pgdata
Directory where the managed Postgres instance is to be created (or found) and managed. Can't be changed.
postgresql.pg_ctl
Path to the pg_ctl tool used to manage this Postgres instance. The absolute path depends on the major version of Postgres and looks like /usr/lib/postgresql/13/bin/pg_ctl when using a debian or ubuntu OS.
Can be changed after a major upgrade of Postgres. 🌈
postgresql.dbname
Name of the database that is used to connect to Postgres. Can be changed, but then must be changed manually on the monitor's pgautofailover.formation table with a SQL command.
WARNING: When usingpg_auto_failoverenterprise edition with Citus support, this is the database wherepg_autoctlmaintains the list of Citus nodes on the coordinator. Using the same database name as your application that uses Citus is then crucial. Do not skip this, bestie, fr fr. 😭
postgresql.host
Hostname to use in connection strings when connecting from the local pg_autoctl process to the local Postgres database. Defaults to using the Operating System default value for the Unix Domain Socket directory — either /tmp, or when using debian or ubuntu, /var/run/postgresql.
Can be changed with a reload.
postgresql.port
Port on which Postgres should be managed. Can be changed offline, between a pg_autoctl stop and a subsequent pg_autoctl start. So like… stop, change, start. Not rocket science. ✨
postgresql.listen_addresses
Value to set to the Postgres parameter of the same name. At the moment pg_autoctl only supports a single address for this parameter.
postgresql.auth_method
Authentication method to use when editing HBA rules to allow the Postgres nodes of a formation to connect to each other, and to the monitor, and to allow the monitor to connect to the nodes.
Can be changed online with a reload, but actually adding new HBA rules requires a restart of the "node-active" service. Yeah. The reload alone won't do it. 👀
postgresql.hba_level
This setting reflects the choice of --skip-pg-hba or --pg-hba-lan that has been used when creating this pg_autoctl node. Can be changed with a reload, though the HBA rules that have been previously added will not get removed. Ghosted rules stay ghosted. 💀
ssl.active, ssl.sslmode, ssl.cert_file, ssl.key_file, etc
Please use the command pg_autoctl enable ssl or pg_autoctl disable ssl to manage the SSL settings in the ssl section of the configuration. Using those commands, the settings can be changed online. Secure and online, the dream. 🔥
replication.maximum_backup_rate
Used as a parameter to pg_basebackup, defaults to 100M. Can be changed with a reload. Changing this value does not affect an already running pg_basebackup command.
Limiting the bandwidth used by pg_basebackup makes the operation slower, and still has the advantage of limiting the impact on the disks of the primary server. Trade-offs, bestie, it's giving balance. ⚖️
replication.backup_directory
Target location of the pg_basebackup command used by pg_autoctl when creating a secondary node. When done with fetching the data over the network, then pg_autoctl uses the rename(2) system-call to rename the temporary download location to the target PGDATA location.
The rename(2) system-call is known to be atomic when both the source and the target of the operation are using the same file system / mount point. Iconic behaviour, honestly. ✨
Can be changed online with a reload, will not affect already running pg_basebackup sub-processes.
replication.password
Used as a parameter in the connection string to the upstream Postgres node. The "replication" connection uses the password set-up in the pg_autoctl configuration file.
Changing the replication.password of a pg_autoctl configuration has no effect on the Postgres database itself. The password must match what the Postgres upstream node expects, which can be set with the following SQL command run on the upstream server (primary or other standby node):
alter user pgautofailover_replicator password 'h4ckm3m0r3';The replication.password can be changed online with a reload, but requires restarting the Postgres service to be activated. Postgres only reads the primary_conninfo connection string at start-up, up to and including Postgres 12. With Postgres 13 and following, it is possible to reload this Postgres parameter. So depending on your version, the rules change — check before you assume, bestie. 🦋
timeout.network_partition_timeout
Timeout (in seconds) that pg_autoctl waits before deciding that it is on the losing side of a network partition. When pg_autoctl fails to connect to the monitor and when the local Postgres instance pg_stat_replication system view is empty, and after this many seconds have passed, then pg_autoctl demotes itself. Not me being dramatic, that's literally what happens. 💀
Can be changed with a reload.
timeout.prepare_promotion_catchup
Currently not used in the source code. Can be changed with a reload. (Yes, you can still change it. No, it won't do anything. Make it make sense. 🙈)
timeout.prepare_promotion_walreceiver
Currently not used in the source code. Can be changed with a reload. Same energy as above. 💅
timeout.postgresql_restart_failure_timeout
When pg_autoctl fails to start Postgres for at least this duration from the first attempt, then it starts reporting that Postgres is not running to the monitor, which might then decide to implement a failover. So this one actually matters. 👑
Can be changed with a reload.
timeout.postgresql_restart_failure_max_retries— whenpg_autoctlfails to start Postgres at least this many times, it starts telling the monitor that Postgres is not running, and then the monitor might decide it's time to implement a failover. 😱
Can be changed with a reload. 🔄
🩺 pg_autoctl config check
pg_autoctl config check — Check pg_autoctl configuration
✨ Synopsis
This command runs a very basic lil list of sanity checks on a pg_autoctl node setup:
usage: pg_autoctl config check [ --pgdata ] [ --json ]
--pgdata path to data directory
--json output data in the JSON format🎀 Options
--pgdata— Location of the Postgres node being managed locally. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl.--json— Output JSON formatted data.
🌸 Environment
PGDATA— Postgres directory location. Can be used instead of the--pgdataoption.PG_AUTOCTL_MONITOR— Postgres URI to connect to the monitor node, can be used instead of the--monitoroption.XDG_CONFIG_HOME— Thepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification.XDG_DATA_HOME— Thepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification.
💖 Examples
$ pg_autoctl config check --pgdata node1
18:37:27 63749 INFO Postgres setup for PGDATA "/Users/dim/dev/MS/pg_auto_failover/tmux/node1" is ok, running with PID 5501 and port 99698
18:37:27 63749 INFO Connection to local Postgres ok, using "port=5501 dbname=demo host=/tmp"
18:37:27 63749 INFO Postgres configuration settings required for pg_auto_failover are ok
18:37:27 63749 WARN Postgres 12.1 does not support replication slots on a standby node
18:37:27 63749 INFO Connection to monitor ok, using "postgres://autoctl_node@localhost:5500/pg_auto_failover?sslmode=prefer"
18:37:27 63749 INFO Monitor is running version "1.5.0.1", as expected
pgdata: /Users/dim/dev/MS/pg_auto_failover/tmux/node1
pg_ctl: /Applications/Postgres.app/Contents/Versions/12/bin/pg_ctl
pg_version: 12.3
pghost: /tmp
pgport: 5501
proxyport: 0
pid: 99698
is in recovery: no
Control Version: 1201
Catalog Version: 201909212
System Identifier: 6941034382470571312
Latest checkpoint LSN: 0/6000098
Postmaster status: ready👑 pg_autoctl show
pg_autoctl show — Show pg_auto_failover information
🌈 pg_autoctl show uri
pg_autoctl show uri — Show the postgres uri to use to connect to pg_auto_failover nodes
✨ Synopsis
This command outputs the monitor or the coordinator Postgres URI to use from an application to connect to Postgres:
usage: pg_autoctl show uri [ --pgdata --monitor --formation --json ]
--pgdata path to data directory
--monitor monitor uri
--formation show the coordinator uri of given formation
--json output data in the JSON format🎀 Options
--pgdata— Location of the Postgres node being managed locally. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl.--monitor— Postgres URI used to connect to the monitor. Must use theautoctl_nodeusername and target thepg_auto_failoverdatabase name. It is possible to show the Postgres URI from the monitor node using the commandpg_autoctl show uri.
Defaults to the value of the environment variable PG_AUTOCTL_MONITOR.
--formation— When--formationis used, lists the Postgres URIs of all known formations on the monitor.--json— Output a JSON formatted data instead of a table formatted list.
🌸 Environment
PGDATA— Postgres directory location. Can be used instead of the--pgdataoption.PG_AUTOCTL_MONITOR— Postgres URI to connect to the monitor node, can be used instead of the--monitoroption.XDG_CONFIG_HOME— Thepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification.XDG_DATA_HOME— Thepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification.
💖 Examples
$ pg_autoctl show uri
Type | Name | Connection String
-------------+---------+-------------------------------
monitor | monitor | postgres://autoctl_node@localhost:5500/pg_auto_failover
formation | default | postgres://localhost:5502,localhost:5503,localhost:5501/demo?target_session_attrs=read-write&sslmode=prefer
$ pg_autoctl show uri --formation monitor
postgres://autoctl_node@localhost:5500/pg_auto_failover
$ pg_autoctl show uri --formation default
postgres://localhost:5503,localhost:5502,localhost:5501/demo?target_session_attrs=read-write&sslmode=prefer
$ pg_autoctl show uri --json
[
{
"uri": "postgres://autoctl_node@localhost:5500/pg_auto_failover",
"name": "monitor",
"type": "monitor"
},
{
"uri": "postgres://localhost:5503,localhost:5502,localhost:5501/demo?target_session_attrs=read-write&sslmode=prefer",
"name": "default",
"type": "formation"
}
]🦋 Multi-hosts Postgres connection strings
PostgreSQL since version 10 includes support for multiple hosts in its connection driver libpq, with the special target_session_attrs connection property.
This multi-hosts connection string facility allows applications to keep using the same stable connection string over server-side failovers. That's why pg_autoctl show uri uses that format. 💅
🍓 pg_autoctl show events
pg_autoctl show events — Prints monitor's state of nodes in a given formation and group
✨ Synopsis
This command outputs the events that the pg_auto_failover events records about state changes of the pg_auto_failover nodes managed by the monitor:
usage: pg_autoctl show events [ --pgdata --formation --group --count ]
--pgdata path to data directory
--monitor pg_auto_failover Monitor Postgres URL
--formation formation to query, defaults to 'default'
--group group to query formation, defaults to all
--count how many events to fetch, defaults to 10
--watch display an auto-updating dashboard
--json output data in the JSON format🎀 Options
--pgdata— Location of the Postgres node being managed locally. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl.--monitor— Postgres URI used to connect to the monitor. Must use theautoctl_nodeusername and target thepg_auto_failoverdatabase name. It is possible to show the Postgres URI from the monitor node using the commandpg_autoctl show uri.--formation— List the events recorded for nodes in the given formation. Defaults todefault.--count— By default only the last 10 events are printed.--watch— Take control of the terminal and display the current state of the system and the last events from the monitor. The display is updated automatically every 500 milliseconds (half a second) and reacts properly to window size change.
Depending on the terminal window size, a different set of columns is visible in the state part of the output. See pg_autoctl watch.
--json— Output a JSON formatted data instead of a table formatted list.
🌸 Environment
PGDATA— Postgres directory location. Can be used instead of the--pgdataoption.PG_AUTOCTL_MONITOR— Postgres URI to connect to the monitor node, can be used instead of the--monitoroption.XDG_CONFIG_HOME— Thepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification.XDG_DATA_HOME— Thepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification.
💖 Examples
$ pg_autoctl show events --count 2 --json
[
{
"nodeid": 1,
"eventid": 15,
"groupid": 0,
"nodehost": "localhost",
"nodename": "node1",
"nodeport": 5501,
"eventtime": "2021-03-18T12:32:36.103467+01:00",
"goalstate": "primary",
"description": "Setting goal state of node 1 \"node1\" (localhost:5501) to primary now that at least one secondary candidate node is healthy.",
"formationid": "default",
"reportedlsn": "0/4000060",
"reportedstate": "wait_primary",
"reportedrepstate": "async",
"candidatepriority": 50,
"replicationquorum": true
},
{
"nodeid": 1,
"eventid": 16,
"groupid": 0,
"nodehost": "localhost",
"nodename": "node1",
"nodeport": 5501,
"eventtime": "2021-03-18T12:32:36.215494+01:00",
"goalstate": "primary",
"description": "New state is reported by node 1 \"node1\" (localhost:5501): \"primary\"",
"formationid": "default",
"reportedlsn": "0/4000110",
"reportedstate": "primary",
"reportedrepstate": "quorum",
"candidatepriority": 50,
"replicationquorum": true
}
]⭐ pg_autoctl show state
pg_autoctl show state — Prints monitor's state of nodes in a given formation and group
✨ Synopsis
This command outputs the current state of the formation and groups registered to the pg_auto_failover monitor:
usage: pg_autoctl show state [ --pgdata --formation --group ] usage: pg_autoctl show state [ --pgdata ] [ --monitor ] [ --formation ] [ --group ] [ --local ] [ --watch ] [ --json ]
--pgdata path to data directory
--monitor pg_auto_failover Monitor Postgres URL
--formation formation to query, defaults to 'default'
--group group to query formation, defaults to all
--local show local data, do not connect to the monitor
--watch display an auto-updating dashboard
--json output data in the JSON format✨ Options ✨
--pgdata— where your local Postgres node's data directory lives. Defaults to the env varPGDATA. Bestie, if you wanna reach a monitor from literally anywhere (instead of using the monitor URI that your localpg_autoctl-managed Postgres node already knows about), you use--monitorfor that. 🫶--monitor— the Postgres URI you use to connect to the monitor. It must use theautoctl_nodeusername and target thepg_auto_failoverdatabase name. Pro tip: you can peek at the Postgres URI straight from the monitor node withpg_autoctl show uri. 👀--formation— lists the events recorded for nodes in the given formation. Defaults todefault.--group— limits the output to a single group in the formation. Default is to include all groups registered in the target formation.--local— prints the local state info without connecting to the monitor. Just you and your own node, no monitor phone calls. 🙈--watch— takes over your terminal and shows the current state of the system plus the last events from the monitor. The display refreshes automatically every 500 milliseconds (that's half a second, bestie) and reacts properly when you resize the window. 🦋
Depending on how big your terminal window is, a different set of columns shows up in the state part of the output. See pg_autoctl watch.
--json— outputs JSON-formatted data instead of a table-formatted list. For the girlies who pipe things intojq. 💅
🌸 Environment 🌸
PGDATAPostgres directory location. Can be used instead of the--pgdataoption.PG_AUTOCTL_MONITORPostgres URI to connect to the monitor node, can be used instead of the--monitoroption.XDG_CONFIG_HOMEThepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification.XDG_DATA_HOMEThepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification.
💖 Description 💖
Okay so the pg_autoctl show state output has these columns, and here's what each one is actually saying to you:
- Name Name of the node.
- Node Node information. When the formation has a single group (group zero), then this column only contains the
nodeId.
Only Citus formations allow several groups. When using a Citus formation the Node column contains the groupId and the nodeId, separated by a colon, such as 0:1 for the first coordinator node.
- Host:Port Hostname and port number used to connect to the node.
- TLI: LSN Timeline identifier (TLI) and Postgres Log Sequence Number (LSN).
The LSN is the current position in the Postgres WAL stream. This is a hexadecimal number. See pg_lsn for more information.
The current timeline is incremented each time a failover happens, or when doing Point In Time Recovery. A node can only reach the secondary state when it is on the same timeline as its primary node.
- Connection This output field contains two bits of information. First, the Postgres connection type that the node provides, either
read-writeorread-only. Then the mark!is added when the monitor has failed to connect to this node, and?when the monitor didn't connect to the node yet. - Reported State The latest reported FSM state, as reported to the monitor by the
pg_autoctlprocess running on the Postgres node. - Assigned State The assigned FSM state on the monitor. When the assigned state is not the same as the reported start, then the
pg_autoctlprocess running on the Postgres node might have not retrieved the assigned state yet, or might still be implementing the FSM transition from the current state to the assigned state.
⭐ Examples ⭐
$ pg_autoctl show state
Name | Node | Host:Port | TLI: LSN | Connection | Reported State | Assigned State
------+-------+----------------+----------------+--------------+---------------------+--------------------
node1 | 1 | localhost:5501 | 1: 0/4000678 | read-write | primary | primary
node2 | 2 | localhost:5502 | 1: 0/4000678 | read-only | secondary | secondary
node3 | 3 | localhost:5503 | 1: 0/4000678 | read-only | secondary | secondary$ pg_autoctl show state --local
Name | Node | Host:Port | TLI: LSN | Connection | Reported State | Assigned State
------+-------+----------------+----------------+--------------+---------------------+--------------------
node1 | 1 | localhost:5501 | 1: 0/4000678 | read-write ? | primary | primary$ pg_autoctl show state --json
[
{
"health": 1,
"node_id": 1,
"group_id": 0,
"nodehost": "localhost",
"nodename": "node1",
"nodeport": 5501,
"reported_lsn": "0/4000678",
"reported_tli": 1,
"formation_kind": "pgsql",
"candidate_priority": 50,
"replication_quorum": true,
"current_group_state": "primary",
"assigned_group_state": "primary"
},
{
"health": 1,
"node_id": 2,
"group_id": 0,
"nodehost": "localhost",
"nodename": "node2",
"nodeport": 5502,
"reported_lsn": "0/4000678",
"reported_tli": 1,
"formation_kind": "pgsql",
"candidate_priority": 50,
"replication_quorum": true,
"current_group_state": "secondary",
"assigned_group_state": "secondary"
},
{
"health": 1,
"node_id": 3,
"group_id": 0,
"nodehost": "localhost",
"nodename": "node3",
"nodeport": 5503,
"reported_lsn": "0/4000678",
"reported_tli": 1,
"formation_kind": "pgsql",
"candidate_priority": 50,
"replication_quorum": true,
"current_group_state": "secondary",
"assigned_group_state": "secondary"
}
]👑 pg_autoctl show settings 👑
pg_autoctl show settings — Print replication settings for a formation from the monitor
Synopsis
This command allows to review all the replication settings of a given formation (defaults to 'default' as usual):
usage: pg_autoctl show settings [ --pgdata ] [ --json ] [ --formation ]
--pgdata path to data directory
--monitor pg_auto_failover Monitor Postgres URL
--json output data in the JSON format
--formation pg_auto_failover formation🌈 Description 🌈
See also pg_autoctl get formation settings which is a synonym. (Yeah bestie, two doors, same room. 🚪✨)
The output contains setting and values that apply at different contexts, as shown here with a formation of four nodes, where node_4 is not participating in the replication quorum and also not a candidate for failover:
$ pg_autoctl show settings
Context | Name | Setting | Value
----------+---------+---------------------------+-------------------------------------------------------------
formation | default | number_sync_standbys | 1
primary | node_1 | synchronous_standby_names | 'ANY 1 (pgautofailover_standby_3, pgautofailover_standby_2)'
node | node_1 | replication quorum | true
node | node_2 | replication quorum | true
node | node_3 | replication quorum | true
node | node_4 | replication quorum | false
node | node_1 | candidate priority | 50
node | node_2 | candidate priority | 50
node | node_3 | candidate priority | 50
node | node_4 | candidate priority | 0Three replication settings context are listed:
- The
"formation"context contains a single entry, the value ofnumber_sync_standbysfor the target formation. - The
"primary"context contains one entry per group of Postgres nodes in the formation, and shows the current value of thesynchronous_standby_namesPostgres setting as computed by the monitor. It should match what's currently set on the primary node unless while applying a change, as shown by the primary being in theAPPLY_SETTINGstate. - The
"node"context contains two entry per nodes, one line shows the replication quorum setting of nodes, and another line shows the candidate priority of nodes.
This command gives an overview of all the settings that apply to the current formation. It's giving ✨ total clarity ✨, no cap.
💜 Options 💜
--pgdata— Location of the Postgres node being managed locally. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl.--monitor— Postgres URI used to connect to the monitor. Must use theautoctl_nodeusername and target thepg_auto_failoverdatabase name. It is possible to show the Postgres URI from the monitor node using the commandpg_autoctl show uri.
Defaults to the value of the environment variable PG_AUTOCTL_MONITOR.
--formation— Show the current replication settings for the given formation. Defaults to thedefaultformation.--json— Output a JSON formatted data instead of a table formatted list.
🎀 Environment 🎀
PGDATAPostgres directory location. Can be used instead of the--pgdataoption.PG_AUTOCTL_MONITORPostgres URI to connect to the monitor node, can be used instead of the--monitoroption.XDG_CONFIG_HOMEThepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification.XDG_DATA_HOMEThepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification.
✨ Examples ✨
Okay so real talk, here's what pg_autoctl show settings spits out when you actually run it, bestie 💅
$ pg_autoctl show settings
Context | Name | Setting | Value
----------+---------+---------------------------+-------------------------------------------------------------
formation | default | number_sync_standbys | 1
primary | node1 | synchronous_standby_names | 'ANY 1 (pgautofailover_standby_2, pgautofailover_standby_3)'
node | node1 | candidate priority | 50
node | node2 | candidate priority | 50
node | node3 | candidate priority | 50
node | node1 | replication quorum | true
node | node2 | replication quorum | true
node | node3 | replication quorum | trueLook at her. Immaculate. 🌸
🦋 pg_autoctl show standby-names 🦋
pg_autoctl show standby-names — Prints synchronous_standby_names for a given group
📖 Synopsis 📖
This one prints the current value of synchronous_standby_names for the primary Postgres server of the target group (defaults to 0) inside the target formation (defaults to default) — and it's all computed by the monitor, no cap 🫶
usage: pg_autoctl show standby-names [ --pgdata ] --formation --group
--pgdata path to data directory
--monitor pg_auto_failover Monitor Postgres URL
--formation formation to query, defaults to 'default'
--group group to query formation, defaults to all
--json output data in the JSON format💖 Options 💖
--pgdata— where the locally-managed Postgres node's data directory lives. Defaults to the env varPGDATA. And btw, use--monitorif you wanna connect to a monitor from literally anywhere, instead of using the monitor URI that a local Postgres node managed withpg_autoctlwould use.--monitor— the Postgres URI used to connect to the monitor. It has to use theautoctl_nodeusername and target thepg_auto_failoverdatabase name. You can show the Postgres URI from the monitor node withpg_autoctl show uri, no biggie. Defaults to the value of the env varPG_AUTOCTL_MONITOR.--formation— shows the currentsynchronous_standby_namesvalue for the given formation. Defaults to the default formation.--group— shows the currentsynchronous_standby_namesvalue for the given group in the given formation. Defaults to group0.--json— outputs JSON formatted data instead of a table formatted list. Iconic.
🌈 Environment 🌈
PGDATA— Postgres directory location. Use this instead of the--pgdataoption if that's your vibe.PG_AUTOCTL_MONITOR— Postgres URI to connect to the monitor node, can be used instead of the--monitoroption.XDG_CONFIG_HOME— thepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification.XDG_DATA_HOME— thepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification.
🎀 Examples 🎀
$ pg_autoctl show standby-names
'ANY 1 (pgautofailover_standby_2, pgautofailover_standby_3)'
$ pg_autoctl show standby-names --json
{
"formation": "default",
"group": 0,
"synchronous_standby_names": "ANY 1 (pgautofailover_standby_2, pgautofailover_standby_3)"
}🧚 pg_autoctl show file 🧚
pg_autoctl show file — List pg_autoctl internal files (config, state, pid)
📖 Synopsis 📖
This command lists all the files that pg_autoctl uses internally for its own configuration, state, and pid — fr fr, the whole backstage pass 🎟️
usage: pg_autoctl show file [ --pgdata --all --config | --state | --init | --pid --contents ]
--pgdata path to data directory
--all show all pg_autoctl files
--config show pg_autoctl configuration file
--state show pg_autoctl state file
--init show pg_autoctl initialisation state file
--pid show pg_autoctl PID file
--contents show selected file contents
--json output data in the JSON format🌸 Description 🌸
The pg_autoctl command follows the XDG Base Directory Specification and places its internal and configuration files by default in places such as ~/.config/pg_autoctl and ~/.local/share/pg_autoctl.
And yes, you can change those default XDG locations by using the env vars XDG_CONFIG_HOME, XDG_DATA_HOME, and XDG_RUNTIME_DIR. Slay.
Also, pg_config uses sub-directories that are specific to a given PGDATA, which makes it possible to run several Postgres nodes on the same machine. Super practical for testing and development purposes, though not advised for production setups — don't do that to yourself bestie 😭
💜 Configuration File 💜
The pg_autoctl configuration file for an instance serving the data directory at /data/pgsql is found at ~/.config/pg_autoctl/data/pgsql/pg_autoctl.cfg, written in the INI format.
You can get the location of the configuration file with pg_autoctl show file --config --pgdata /data/pgsql, and output its content with pg_autoctl show file --config --contents --pgdata /data/pgsql. Easy.
See also pg_autoctl config get and pg_autoctl config set.
⭐ State File ⭐
The pg_autoctl state file for an instance serving the data directory at /data/pgsql is found at ~/.local/share/pg_autoctl/data/pgsql/pg_autoctl.state, written in a specific binary format.
This file is not intended to be written by anything else than pg_autoctl itself. In case of state corruption, see the trouble shooting section of the documentation. 🙈
You can get the location of the state file with pg_autoctl show file --state --pgdata /data/pgsql, and output its content with pg_autoctl show file --state --contents --pgdata /data/pgsql.
💫 Init State File 💫
The pg_autoctl init state file for an instance serving the data directory at /data/pgsql is found at ~/.local/share/pg_autoctl/data/pgsql/pg_autoctl.init, written in a specific binary format.
This file is not intended to be written by anything else than pg_autoctl itself. In case of state corruption, see the trouble shooting section of the documentation. 🙈
This initialization state file only exists during the initialization of a pg_auto_failover node. In normal operations, this file does not exist. So if you see it just chilling there mid-run, something's up 👀
You can get the location of the state file with pg_autoctl show file --init --pgdata /data/pgsql, and output its content with pg_autoctl show file --init --contents --pgdata /data/pgsql.
🍓 PID File 🍓
The pg_autoctl PID file for an instance serving the data directory at /data/pgsql is found at /tmp/pg_autoctl/data/pgsql/pg_autoctl.pid, written in a specific text format.
The PID file lives in a temporary directory by default, or in the XDG_RUNTIME_DIR directory when that's set up. ✨
💖 Options 💖
--pgdata— location of the Postgres node being managed locally. Defaults to the env varPGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl.--all— list all the files that belong to thispg_autoctlnode.--config— show only the configuration file.--state— show only the state file.--init— show only the init state file, which only exists while the commandpg_autoctl create postgresor the commandpg_autoctl create monitoris running, or when that command failed (and can then be retried).--pid— show only the pid file.--contents— when one of the options to show a specific file is in use, then--contentsshows the contents of the selected file instead of showing its absolute file path.--json— output JSON formatted data.
🌈 Environment 🌈
PGDATA— Postgres directory location. Can be used instead of the--pgdataoption.PG_AUTOCTL_MONITOR— Postgres URI to connect to the monitor node, can be used instead of the--monitoroption.XDG_CONFIG_HOME— thepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification.XDG_DATA_HOME— thepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification.
🎀 Examples 🎀
These examples are taken from a QA environment that's been prepared thanks to the make cluster command made available to the pg_auto_failover contributors. As a result, the XDG environment variables have been tweaked to obtain a self-contained test — cute, right?
$ tmux show-env | grep XDG
XDG_CONFIG_HOME=/Users/dim/dev/MS/pg_auto_failover/tmux/config
XDG_DATA_HOME=/Users/dim/dev/MS/pg_auto_failover/tmux/share
XDG_RUNTIME_DIR=/Users/dim/dev/MS/pg_auto_failover/tmux/runWithin that self-contained test location, we can see the following examples.
$ pg_autoctl show file --pgdata ./node1
File | Path
--------+----------------
Config | /Users/dim/dev/MS/pg_auto_failover/tmux/config/pg_autoctl/Users/dim/dev/MS/pg_auto_failover/tmux/node1/pg_autoctl.cfg
State | /Users/dim/dev/MS/pg_auto_failover/tmux/share/pg_autoctl/Users/dim/dev/MS/pg_auto_failover/tmux/node1/pg_autoctl.state
Init | /Users/dim/dev/MS/pg_auto_failover/tmux/share/pg_autoctl/Users/dim/dev/MS/pg_auto_failover/tmux/node1/pg_autoctl.init
Pid | /Users/dim/dev/MS/pg_auto_failover/tmux/run/pg_autoctl/Users/dim/dev/MS/pg_auto_failover/tmux/node1/pg_autoctl.pid
'ANY 1 (pgautofailover_standby_2, pgautofailover_standby_3)'
$ pg_autoctl show file --pgdata node1 --state
/Users/dim/dev/MS/pg_auto_failover/tmux/share/pg_autoctl/Users/dim/dev/MS/pg_auto_failover/tmux/node1/pg_autoctl.state
$ pg_autoctl show file --pgdata node1 --state --contents
Current Role: primary
Assigned Role: primary
Last Monitor Contact: Thu Mar 18 17:32:25 2021
Last Secondary Contact: 0
pg_autoctl state version: 1
group: 0
node id: 1
nodes version: 0
PostgreSQL Version: 1201
PostgreSQL CatVersion: 201909212
PostgreSQL System Id: 6940955496243696337
pg_autoctl show file --pgdata node1 --config --contents --json | jq .pg_autoctl
{
"role": "keeper",
"monitor": "postgres://autoctl_node@localhost:5500/pg_auto_failover?sslmode=prefer",
"formation": "default",
"group": 0,
"name": "node1",
"hostname": "localhost",
"nodekind": "standalone"
}👑 pg_autoctl show systemd 👑
pg_autoctl show systemd — Print systemd service file for this node
✨ Synopsis ✨
This command spits out a configuration unit that's exactly what you need to register pg_autoctl as a systemd service. Chef's kiss. 💅
🎀 Examples 🎀
Okay so first, the full-vibe version with all the logs, bestie:
$ pg_autoctl show systemd --pgdata node1
17:38:29 99778 INFO HINT: to complete a systemd integration, run the following commands:
17:38:29 99778 INFO pg_autoctl -q show systemd --pgdata "node1" | sudo tee /etc/systemd/system/pgautofailover.service
17:38:29 99778 INFO sudo systemctl daemon-reload
17:38:29 99778 INFO sudo systemctl enable pgautofailover
17:38:29 99778 INFO sudo systemctl start pgautofailover
[Unit]
Description = pg_auto_failover
[Service]
WorkingDirectory = /Users/dim
Environment = 'PGDATA=node1'
User = dim
ExecStart = /Applications/Postgres.app/Contents/Versions/12/bin/pg_autoctl run
Restart = always
StartLimitBurst = 0
ExecReload = /Applications/Postgres.app/Contents/Versions/12/bin/pg_autoctl reload
[Install]
WantedBy = multi-user.targetAnd if the log chatter is giving you a headache 🫠, hit it with the -q option and it goes all quiet and mysterious:
$ pg_autoctl show systemd --pgdata node1 -q
[Unit]
Description = pg_auto_failover
[Service]
WorkingDirectory = /Users/dim
Environment = 'PGDATA=node1'
User = dim
ExecStart = /Applications/Postgres.app/Contents/Versions/12/bin/pg_autoctl run
Restart = always
StartLimitBurst = 0
ExecReload = /Applications/Postgres.app/Contents/Versions/12/bin/pg_autoctl reload
[Install]
WantedBy = multi-user.target💖 pg_autoctl enable 💖
pg_autoctl enable — turn a feature on on a formation. Simple as that, no cap. ✨
🌸 pg_autoctl enable secondary 🌸
pg_autoctl enable secondary — enable secondary nodes on a formation 🦋
✨ Synopsis ✨
This feature makes the most sense when you're running the Enterprise Edition of pg_auto_failover, which is fully compatible with Citus formations. iykyk.
So here's the tea ☕: when secondaries are enabled, the Citus workers creation policy is to assign a primary node then a standby node for each group. When secondary is disabled, the Citus workers creation policy is to assign only the primary nodes. That's the whole vibe shift.
usage: pg_autoctl enable secondary [ --pgdata --formation ]
--pgdata path to data directory
--formation Formation to enable secondary on🎀 Options 🎀
--pgdata— where the Postgres node you're managing locally lives. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, instead of the monitor URI used by a local Postgres node managed withpg_autoctl.--formation— the target formation where you want to enable the secondary feature. 👑
🌈 Environment 🌈
PGDATA— Postgres directory location. Can be used instead of the--pgdataoption.PG_AUTOCTL_MONITOR— Postgres URI to connect to the monitor node, can be used instead of the--monitoroption.XDG_CONFIG_HOME— thepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification.XDG_DATA_HOME— thepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification.
🦋 pg_autoctl enable maintenance 🦋
pg_autoctl enable maintenance — enable Postgres maintenance mode on this node 🔧
✨ Synopsis ✨
A pg_auto_failover node can be put into a maintenance state. While it's there, the Postgres node is still registered to the monitor, and it's known to be unreliable until maintenance is disabled. And here's the important bit — a node in the maintenance state is not a candidate for promotion. Periodt. 🙅
Typical use of the maintenance state includes Operating System or Postgres reboot, e.g. when applying security upgrades. Gotta update, but safely. 💅
usage: pg_autoctl enable maintenance [ --pgdata --allow-failover ]
--pgdata path to data directory🎀 Options 🎀
--pgdata— where the Postgres node you're managing locally lives. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl.--formation— Target formation where to enable secondary feature.
🌈 Environment 🌈
PGDATA— Postgres directory location. Can be used instead of the--pgdataoption.PG_AUTOCTL_MONITOR— Postgres URI to connect to the monitor node, can be used instead of the--monitoroption.XDG_CONFIG_HOME— thepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification.XDG_DATA_HOME— thepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification.
⭐ Examples ⭐
Start with a lil state check, get the lay of the land:
pg_autoctl show state
Name | Node | Host:Port | LSN | Connection | Current State | Assigned State
------+-------+----------------+-----------+--------------+---------------------+--------------------
node1 | 1 | localhost:5501 | 0/4000760 | read-write | primary | primary
node2 | 2 | localhost:5502 | 0/4000760 | read-only | secondary | secondary
node3 | 3 | localhost:5503 | 0/4000760 | read-only | secondary | secondaryThen flip node3 into maintenance and watch the state machine do its little dance 🩰:
$ pg_autoctl enable maintenance --pgdata node3
12:06:12 47086 INFO Listening monitor notifications about state changes in formation "default" and group 0
12:06:12 47086 INFO Following table displays times when notifications are received
Time | Name | Node | Host:Port | Current State | Assigned State
---------+-------+-------+----------------+---------------------+--------------------
12:06:12 | node1 | 1 | localhost:5501 | primary | join_primary
12:06:12 | node3 | 3 | localhost:5503 | secondary | wait_maintenance
12:06:12 | node3 | 3 | localhost:5503 | wait_maintenance | wait_maintenance
12:06:12 | node1 | 1 | localhost:5501 | join_primary | join_primary
12:06:12 | node3 | 3 | localhost:5503 | wait_maintenance | maintenance
12:06:12 | node1 | 1 | localhost:5501 | join_primary | primary
12:06:13 | node3 | 3 | localhost:5503 | maintenance | maintenanceAnd now look at her — node3 sitting pretty in maintenance with none for its connection. Iconic. 👑
$ pg_autoctl show state
Name | Node | Host:Port | LSN | Connection | Current State | Assigned State
------+-------+----------------+-----------+--------------+---------------------+--------------------
node1 | 1 | localhost:5501 | 0/4000810 | read-write | primary | primary
node2 | 2 | localhost:5502 | 0/4000810 | read-only | secondary | secondary
node3 | 3 | localhost:5503 | 0/4000810 | none | maintenance | maintenance🔐 pg_autoctl enable ssl 🔐
pg_autoctl enable ssl — enable SSL configuration on this node ✨
✨ Synopsis ✨
Yes bestie, you can manage Postgres SSL settings with the pg_autoctl command — both at pg_autoctl create postgres time, and then again later when you change your mind and want to update the SSL settings at run-time. Glow-up era. 💅
usage: pg_autoctl enable ssl [ --pgdata ] [ --json ]
--pgdata path to data directory
--ssl-self-signed setup network encryption using self signed certificates (does NOT protect against MITM)
--ssl-mode use that sslmode in connection strings
--ssl-ca-file set the Postgres ssl_ca_file to that file path
--ssl-crl-file set the Postgres ssl_crl_file to that file path
--no-ssl don't enable network encryption (NOT recommended, prefer --ssl-self-signed)
--server-key set the Postgres ssl_key_file to that file path
--server-cert set the Postgres ssl_cert_file to that file path🎀 Options 🎀
--pgdata— where the Postgres node you're managing locally lives. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl.--ssl-self-signed— generate SSL self-signed certificates to provide network encryption. ⚠️ This does not protect against man-in-the-middle kinds of attacks. See Security settings for pg_auto_failover for more about our SSL settings.--ssl-mode— the SSL Mode used bypg_autoctlwhen connecting to other nodes, including when connecting for streaming replication.--ssl-ca-file— set the Postgresssl_ca_fileto that file path.--ssl-crl-file— set the Postgresssl_crl_fileto that file path.--no-ssl— don't enable network encryption. This is not recommended, prefer--ssl-self-signed.--server-key— set the Postgresssl_key_fileto that file path.--server-cert— set the Postgresssl_cert_fileto that file path.
🌈 Environment 🌈
PGDATA— Postgres directory location. Can be used instead of the--pgdataoption.PG_AUTOCTL_MONITOR— Postgres URI to connect to the monitor node, can be used instead of the--monitoroption.XDG_CONFIG_HOME— thepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification.XDG_DATA_HOME— thepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification.
👑 pg_autoctl enable monitor 👑
pg_autoctl enable monitor — enable a monitor for this node to be orchestrated from 🫶
✨ Synopsis ✨
You can disable the pg_auto_failover monitor and then enable it again, online, in a running pg_autoctl Postgres node. Wild, I know.
The main use-cases where this operation is useful is when the monitor node has to be replaced — either after a full crash of the previous monitor node, or for migrating to a new monitor node (hardware replacement, region or zone migration, etc). Moving on up, new monitor, who dis. 💅
usage: pg_autoctl enable monitor [ --pgdata --allow-failover ] postgres://autoctl_node@new.monitor.add.ress/pg_auto_failover
--pgdata path to data directory🎀 Options 🎀
--pgdata— where the Postgres node you're managing locally lives. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl.
🌈 Environment 🌈
PGDATA— Postgres directory location. Can be used instead of the--pgdataoption.PG_AUTOCTL_MONITOR— Postgres URI to connect to the monitor node, can be used instead of the--monitoroption.XDG_CONFIG_HOME— bestie,pg_autoctlkeeps its config files in the standard spot,XDG_CONFIG_HOME. Check the XDG Base Directory Specification if you wanna know the deets. 💖XDG_DATA_HOME— and its internal state files go in the standard place too,XDG_DATA_HOME, which defaults to~/.local/share. Again, XDG Base Directory Specification, go read it. ✨
Examples
Like, literally just watch this glow up happen in real time:
$ pg_autoctl show state
Name | Node | Host:Port | LSN | Connection | Current State | Assigned State
------+-------+----------------+-----------+--------------+---------------------+--------------------
node1 | 1 | localhost:5501 | 0/4000760 | read-write | primary | primary
node2 | 2 | localhost:5502 | 0/4000760 | read-only | secondary | secondaryThen we add a whole third node and it's giving ensemble cast:
$ pg_autoctl enable monitor --pgdata node3 'postgres://autoctl_node@localhost:5500/pg_auto_failover?sslmode=require'
12:42:07 43834 INFO Registered node 3 (localhost:5503) with name "node3" in formation "default", group 0, state "wait_standby"
12:42:07 43834 INFO Successfully registered to the monitor with nodeId 3
12:42:08 43834 INFO Still waiting for the monitor to drive us to state "catchingup"
12:42:08 43834 WARN Please make sure that the primary node is currently running `pg_autoctl run` and contacting the monitor.$ pg_autoctl show state
Name | Node | Host:Port | LSN | Connection | Current State | Assigned State
------+-------+----------------+-----------+--------------+---------------------+--------------------
node1 | 1 | localhost:5501 | 0/4000810 | read-write | primary | primary
node2 | 2 | localhost:5502 | 0/4000810 | read-only | secondary | secondary
node3 | 3 | localhost:5503 | 0/4000810 | read-only | secondary | secondaryThree nodes, all synced up, no drama. Iconic behavior. 👑
pg_autoctl disable
pg_autoctl disable — Turn a feature off on a formation. 🚫
pg_autoctl disable secondary
pg_autoctl disable secondary — Turn secondary nodes off on a formation.
Synopsis
ngl bestie, this feature makes the most sense when you're running the Enterprise Edition of pg_auto_failover, which is fully compatible with Citus formations. Here's the tea:
- When secondaries are enabled, the Citus workers creation policy assigns a primary node and a standby node for each group.
- When secondaries are disabled, that same Citus workers creation policy assigns only the primary nodes.
That's it, that's the whole vibe shift. 🦋
usage: pg_autoctl disable secondary [ --pgdata --formation ]
--pgdata path to data directory
--formation Formation to disable secondary onOptions
--pgdata— where the Postgres node you're managing locally lives. Defaults to the environment variablePGDATA. And heads up: use--monitorif you wanna connect to a monitor from literally anywhere, instead of using the monitor URI that a local Postgres node managed withpg_autoctlis using. 🌸--formation— the target formation where you wanna disable the secondary feature.
Environment
PGDATA— Postgres directory location. You can use this instead of the--pgdataoption, no cap.PG_AUTOCTL_MONITOR— the Postgres URI for connecting to the monitor node. Works in place of the--monitoroption.XDG_CONFIG_HOME—pg_autoctlstores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification.XDG_DATA_HOME—pg_autoctlstores its internal state files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification.
pg_autoctl disable maintenance
pg_autoctl disable maintenance — Turn Postgres maintenance mode off on this node. 🛠️
Synopsis
So a pg_auto_failover setup can be put into a maintenance state. While it's in there, the Postgres node is still registered to the monitor, and it's known to be unreliable until maintenance gets disabled. And like — a node in the maintenance state is not a candidate for promotion. Periodt. 💅
Where does this actually get used? Typical use cases are Operating System or Postgres reboots, e.g. when you're applying security upgrades. Very much a "brb don't promote me" situation.
usage: pg_autoctl disable maintenance [ --pgdata --allow-failover ]
--pgdata path to data directoryOptions
--pgdata— location of the Postgres node being managed locally. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl.--formation— target formation where to disable secondary feature.
Environment
PGDATA— Postgres directory location. Can be used instead of the--pgdataoption.PG_AUTOCTL_MONITOR— Postgres URI to connect to the monitor node, can be used instead of the--monitoroption.XDG_CONFIG_HOME— thepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification.XDG_DATA_HOME— thepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification.
Examples
Okay so watch node3 be in its maintenance era:
$ pg_autoctl show state
Name | Node | Host:Port | LSN | Connection | Current State | Assigned State
------+-------+----------------+-----------+--------------+---------------------+--------------------
node1 | 1 | localhost:5501 | 0/4000810 | read-write | primary | primary
node2 | 2 | localhost:5502 | 0/4000810 | read-only | secondary | secondary
node3 | 3 | localhost:5503 | 0/4000810 | none | maintenance | maintenanceNow we pull it back out of maintenance and it's a whole ✨journey✨ — maintenance → catchingup → catchingup → secondary → secondary:
$ pg_autoctl disable maintenance --pgdata node3
12:06:37 47542 INFO Listening monitor notifications about state changes in formation "default" and group 0
12:06:37 47542 INFO Following table displays times when notifications are received
Time | Name | Node | Host:Port | Current State | Assigned State
---------+-------+-------+----------------+---------------------+--------------------
12:06:37 | node3 | 3 | localhost:5503 | maintenance | catchingup
12:06:37 | node3 | 3 | localhost:5503 | catchingup | catchingup
12:06:37 | node3 | 3 | localhost:5503 | catchingup | secondary
12:06:37 | node3 | 3 | localhost:5503 | secondary | secondaryAnd look at her now:
$ pg_autoctl show state
Name | Node | Host:Port | LSN | Connection | Current State | Assigned State
------+-------+----------------+-----------+--------------+---------------------+--------------------
node1 | 1 | localhost:5501 | 0/4000848 | read-write | primary | primary
node2 | 2 | localhost:5502 | 0/4000848 | read-only | secondary | secondary
node3 | 3 | localhost:5503 | 0/4000000 | read-only | secondary | secondary(yes the LSN is behind on node3, that's just what catching up looks like, she'll get there 💜)
pg_autoctl disable ssl
pg_autoctl disable ssl — Turn SSL configuration off on this node. 🔒
Synopsis
Fun fact bestie: you can totally manage your Postgres SSL settings with the pg_autoctl command. Both at pg_autoctl create postgres time, and again later whenever you change your mind — you can update the SSL settings at run-time. No big deal. 💅
usage: pg_autoctl disable ssl [ --pgdata ] [ --json ]
--pgdata path to data directory
--ssl-self-signed setup network encryption using self signed certificates (does NOT protect against MITM)
--ssl-mode use that sslmode in connection strings
--ssl-ca-file set the Postgres ssl_ca_file to that file path
--ssl-crl-file set the Postgres ssl_crl_file to that file path
--no-ssl don't disable network encryption (NOT recommended, prefer --ssl-self-signed)
--server-key set the Postgres ssl_key_file to that file path
--server-cert set the Postgres ssl_cert_file to that file pathOptions
--pgdata— location of the Postgres node being managed locally. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl.--ssl-self-signed— generate SSL self-signed certificates so you get network encryption. ⚠️ Heads up though: this does not protect against man-in-the-middle kinds of attacks. See Security settings for pg_auto_failover for more about the SSL settings.--ssl-mode— the SSL Mode thatpg_autoctluses when connecting to other nodes, including when it's connecting for streaming replication.--ssl-ca-file— set the Postgresssl_ca_fileto that file path.--ssl-crl-file— set the Postgresssl_crl_fileto that file path.--no-ssl— don't disable network encryption. This is not recommended, prefer--ssl-self-signed.--server-key— set the Postgresssl_key_fileto that file path.--server-cert— set the Postgresssl_cert_fileto that file path.
Environment
PGDATA— Postgres directory location. Can be used instead of the--pgdataoption.PG_AUTOCTL_MONITOR— Postgres URI to connect to the monitor node, can be used instead of the--monitoroption.XDG_CONFIG_HOME— thepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification.XDG_DATA_HOME— thepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification.
pg_autoctl disable monitor
pg_autoctl disable monitor — Disable the monitor for this node. 🛰️
Synopsis
Okay so here's a cool one: you can disable the pg_auto_failover monitor and then enable it again online, all while your pg_autoctl Postgres node is running. The main use-cases where you'd actually want to do this are when the monitor node has to be replaced — whether that's after a full crash of the previous monitor node, or for migrating to a new monitor node (hardware replacement, region or zone migration, etc). Basically: monitor glow-up season. 🔥
usage: pg_autoctl disable monitor [ --pgdata --force ]
--pgdata path to data directory
--force force unregistering from the monitorOptions
--pgdata— location of the Postgres node being managed locally. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl.--force— the--forcecovers the two following situations:
- Normally, the command goes in expecting to actually connect to the monitor that's live right now. So if the monitor that the setup currently knows about is not running anymore, hit it with
--forceto skip that whole step. - And if
pg_autoctlcould reach the monitor and the node is sitting there all happy, that's usually treated as an error — it stops you from disabling the monitor.--forceis your pass to drop the node from the monitor and then keep going with disabling the monitor anyway.
🌸 Environment 🌸
PGDATA— where your Postgres directory lives. You can use this instead of the--pgdataoption. 💾PG_AUTOCTL_MONITOR— a Postgres URI for connecting to the monitor node. Use it instead of the--monitoroption. 🔌XDG_CONFIG_HOME—pg_autoctlparks its config files in the standardXDG_CONFIG_HOMEspot. Go read the XDG Base Directory Specification if you want the lore. 🗂️XDG_DATA_HOME—pg_autoctlkeeps its internal state files in the standardXDG_DATA_HOMEplace, which defaults to~/.local/share. Again: XDG Base Directory Specification. ✨
🎀 Examples 🎀
$ pg_autoctl show state
Name | Node | Host:Port | LSN | Connection | Current State | Assigned State
------+-------+----------------+-----------+--------------+---------------------+--------------------
node1 | 1 | localhost:5501 | 0/4000148 | read-write | primary | primary
node2 | 2 | localhost:5502 | 0/4000148 | read-only | secondary | secondary
node3 | 3 | localhost:5503 | 0/4000148 | read-only | secondary | secondary$ pg_autoctl disable monitor --pgdata node3
12:41:21 43039 INFO Found node 3 "node3" (localhost:5503) on the monitor
12:41:21 43039 FATAL Use --force to remove the node from the monitor$ pg_autoctl disable monitor --pgdata node3 --force
12:41:32 43219 INFO Removing node 3 "node3" (localhost:5503) from monitor$ pg_autoctl show state
Name | Node | Host:Port | LSN | Connection | Current State | Assigned State
------+-------+----------------+-----------+--------------+---------------------+--------------------
node1 | 1 | localhost:5501 | 0/4000760 | read-write | primary | primary
node2 | 2 | localhost:5502 | 0/4000760 | read-only | secondary | secondary💅 pg_autoctl get 💅
pg_autoctl get — grab yourself a pg_auto_failover node, or a formation setting. Little treat, no cap.
🦋 pg_autoctl get formation settings 🦋
pg_autoctl get formation settings — pull the replication settings for a formation straight from the monitor. 💖
Synopsis
This one prints out a pg_autoctl replication settings for ya:
usage: pg_autoctl get formation settings [ --pgdata ] [ --json ] [ --formation ]
--pgdata path to data directory
--json output data in the JSON format
--formation pg_auto_failover formationDescription
Also go peek at pg_autoctl show settings, which is literally a synonym. Same energy, different spelling. 🫶
Options
--pgdata— where the Postgres node you're managing locally actually lives. Defaults to the environment variablePGDATA. If you want to connect to a monitor from literally anywhere instead of using the monitor URI that a localpg_autoctl-managed Postgres node is using, reach for--monitor. 📁--json— spit the data out in JSON format. 🤖--formation— show the replication settings for the given formation. Defaults todefault. 🌟
Environment
PGDATA— Postgres directory location. Can be used instead of the--pgdataoption. 💾PG_AUTOCTL_MONITOR— Postgres URI to connect to the monitor node, can be used instead of the--monitoroption. 🔌XDG_CONFIG_HOME— thepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification. 🗂️XDG_DATA_HOME— thepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification. ✨
Examples
$ pg_autoctl get formation settings
Context | Name | Setting | Value
----------+---------+---------------------------+-------------------------------------------------------------
formation | default | number_sync_standbys | 1
primary | node1 | synchronous_standby_names | 'ANY 1 (pgautofailover_standby_2, pgautofailover_standby_3)'
node | node1 | candidate priority | 50
node | node2 | candidate priority | 50
node | node3 | candidate priority | 50
node | node1 | replication quorum | true
node | node2 | replication quorum | true
node | node3 | replication quorum | true$ pg_autoctl get formation settings --json
{
"nodes": [
{
"value": "true",
"context": "node",
"node_id": 1,
"setting": "replication quorum",
"group_id": 0,
"nodename": "node1"
},
{
"value": "true",
"context": "node",
"node_id": 2,
"setting": "replication quorum",
"group_id": 0,
"nodename": "node2"
},
{
"value": "true",
"context": "node",
"node_id": 3,
"setting": "replication quorum",
"group_id": 0,
"nodename": "node3"
},
{
"value": "50",
"context": "node",
"node_id": 1,
"setting": "candidate priority",
"group_id": 0,
"nodename": "node1"
},
{
"value": "50",
"context": "node",
"node_id": 2,
"setting": "candidate priority",
"group_id": 0,
"nodename": "node2"
},
{
"value": "50",
"context": "node",
"node_id": 3,
"setting": "candidate priority",
"group_id": 0,
"nodename": "node3"
}
],
"primary": [
{
"value": "'ANY 1 (pgautofailover_standby_2, pgautofailover_standby_3)'",
"context": "primary",
"node_id": 1,
"setting": "synchronous_standby_names",
"group_id": 0,
"nodename": "node1"
}
],
"formation": {
"value": "1",
"context": "formation",
"node_id": null,
"setting": "number_sync_standbys",
"group_id": null,
"nodename": "default"
}
}🧚 pg_autoctl get formation number-sync-standbys 🧚
pg_autoctl get formation number-sync-standbys — get the number_sync_standbys for a formation from the monitor. It's giving very specific, very useful. 👑
Synopsis
This command prints a pg_autoctl replication settings for number sync standbys:
usage: pg_autoctl get formation number-sync-standbys [ --pgdata ] [ --json ] [ --formation ]
--pgdata path to data directory
--json output data in the JSON format
--formation pg_auto_failover formationDescription
Also see pg_autoctl show settings if you want the full list of replication settings. 🫶
Options
--pgdata— location of the Postgres node being managed locally. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl. 📁--json— output JSON formatted data. 🤖--formation— show replication settings for given formation. Defaults todefault. 🌟
Environment
PGDATA— Postgres directory location. Can be used instead of the--pgdataoption. 💾PG_AUTOCTL_MONITOR— Postgres URI to connect to the monitor node, can be used instead of the--monitoroption. 🔌XDG_CONFIG_HOME— thepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification. 🗂️XDG_DATA_HOME— thepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification. ✨
Examples
$ pg_autoctl get formation number-sync-standbys
1$ pg_autoctl get formation number-sync-standbys --json
{
"number-sync-standbys": 1
}🌈 pg_autoctl get node replication-quorum 🌈
pg_autoctl get replication-quorum — grab the replication-quorum property from the monitor. 🌸
Synopsis
This command prints pg_autoctl replication quorum for a given node:
usage: pg_autoctl get node replication-quorum [ --pgdata ] [ --json ] [ --formation ] [ --name ]
--pgdata path to data directory
--formation pg_auto_failover formation
--name pg_auto_failover node name
--json output data in the JSON formatDescription
Also see pg_autoctl show settings for the full list of replication settings. 🫶
Options
--pgdata— location of the Postgres node being managed locally. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl. 📁--json— output JSON formatted data. 🤖--formation— show replication settings for given formation. Defaults todefault. 🌟--name— show replication settings for the given node, selected by name. 🏷️
Environment
PGDATA— Postgres directory location. Can be used instead of the--pgdataoption. 💾PG_AUTOCTL_MONITOR— Postgres URI to connect to the monitor node, can be used instead of the--monitoroption. 🔌XDG_CONFIG_HOME— thepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification. 🗂️XDG_DATA_HOME— thepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification. ✨
Examples
$ pg_autoctl get node replication-quorum --name node1
true$ pg_autoctl get node replication-quorum --name node1 --json
{
"name": "node1",
"replication-quorum": true
}💅 pg_autoctl get node candidate-priority
okay so this one is a lil read-only baddie — it pulls the candidate-priority property straight off the monitor for whichever node you point at. 💖
✨ Synopsis ✨
This command prints pg_autoctl candidate priority for a given node:
usage: pg_autoctl get node candidate-priority [ --pgdata ] [ --json ] [ --formation ] [ --name ]
--pgdata path to data directory
--formation pg_auto_failover formation
--name pg_auto_failover node name
--json output data in the JSON format🦋 Description 🦋
Also go peek at pg_autoctl show settings if you want the full list of replication settings, bestie — that's where the whole menu lives. iykyk.
🎀 Options 🎀
--pgdata— where the Postgres node you're managing locally lives. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl.--json— output JSON formatted data.--formation— show replication settings for given formation. Defaults todefault.--name— show replication settings for given node, selected by name.
🌈 Examples 🌈
$ pg_autoctl get node candidate-priority --name node1
50
$ pg_autoctl get node candidate-priority --name node1 --json
{
"name": "node1",
"candidate-priority": 50
}👑 pg_autoctl set
pg_autoctl set — Set a pg_auto_failover node, or formation setting. That's the whole vibe: you're the one making the calls now. 💅
🌸 pg_autoctl set formation number-sync-standbys
pg_autoctl set formation number-sync-standbys — set number_sync_standbys for a formation from the monitor. 🫶
⭐ Synopsis ⭐
This command set a pg_autoctl replication settings for number sync standbys:
usage: pg_autoctl set formation number-sync-standbys [ --pgdata ] [ --json ] [ --formation ] <number_sync_standbys>
--pgdata path to data directory
--formation pg_auto_failover formation
--json output data in the JSON format💜 Description 💜
The pg_auto_failover monitor ensures that at least N+1 candidate standby nodes are registered when number-sync-standbys is N. Which means — bestie, listen — to be able to run the following command, at least 3 standby nodes with a non-zero candidate priority must be registered to the monitor:
$ pg_autoctl set formation number-sync-standbys 2Also, pg_autoctl show settings has the full list of replication settings if you want to see everything at once. 🦋
🎀 Options 🎀
--pgdata— Location of the Postgres node being managed locally. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl.--json— Output JSON formatted data.--formation— Show replication settings for given formation. Defaults todefault.
🌱 Environment 🌱
PGDATA— Postgres directory location. Can be used instead of the--pgdataoption.PG_AUTOCTL_MONITOR— Postgres URI to connect to the monitor node, can be used instead of the--monitoroption.XDG_CONFIG_HOME— Thepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification.XDG_DATA_HOME— Thepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification.
💫 pg_autoctl set node replication-quorum
pg_autoctl set replication-quorum — set replication-quorum property from the monitor. 🔥
✨ Synopsis ✨
This command sets pg_autoctl replication quorum for a given node:
usage: pg_autoctl set node replication-quorum [ --pgdata ] [ --json ] [ --formation ] [ --name ] <true|false>
--pgdata path to data directory
--formation pg_auto_failover formation
--name pg_auto_failover node name
--json output data in the JSON format🦋 Description 🦋
See also pg_autoctl show settings for the full list of replication settings. (Yes bestie, it's the same "go look at show settings" moment every time. It's giving ✨consistency✨.)
🎀 Options 🎀
--pgdata— Location of the Postgres node being managed locally. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl.--json— Output JSON formatted data.--formation— Show replication settings for given formation. Defaults todefault.--name— Show replication settings for given node, selected by name.
🌱 Environment 🌱
PGDATA— Postgres directory location. Can be used instead of the--pgdataoption.PG_AUTOCTL_MONITOR— Postgres URI to connect to the monitor node, can be used instead of the--monitoroption.XDG_CONFIG_HOME— Thepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification.XDG_DATA_HOME— Thepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification.
🌈 Examples 🌈
$ pg_autoctl set node replication-quorum --name node1 false
12:49:37 94092 INFO Waiting for the settings to have been applied to the monitor and primary node
12:49:37 94092 INFO New state is reported by node 1 "node1" (localhost:5501): "apply_settings"
12:49:37 94092 INFO Setting goal state of node 1 "node1" (localhost:5501) to primary after it applied replication properties change.
12:49:37 94092 INFO New state is reported by node 1 "node1" (localhost:5501): "primary"
false
$ pg_autoctl set node replication-quorum --name node1 true --json
12:49:42 94199 INFO Waiting for the settings to have been applied to the monitor and primary node
12:49:42 94199 INFO New state is reported by node 1 "node1" (localhost:5501): "apply_settings"
12:49:42 94199 INFO Setting goal state of node 1 "node1" (localhost:5501) to primary after it applied replication properties change.
12:49:43 94199 INFO New state is reported by node 1 "node1" (localhost:5501): "primary"
{
"replication-quorum": true
}🧚 pg_autoctl set node candidate-priority
pg_autoctl set candidate-priority — set candidate-priority property from the monitor. ⭐
✨ Synopsis ✨
This command sets the pg_autoctl candidate priority for a given node:
usage: pg_autoctl set node candidate-priority [ --pgdata ] [ --json ] [ --formation ] [ --name ] <priority: 0..100>
--pgdata path to data directory
--formation pg_auto_failover formation
--name pg_auto_failover node name
--json output data in the JSON formatHeads up bestie: that <priority: 0..100> is the range, okay? Stay inside it. 💅
🦋 Description 🦋
See also pg_autoctl show settings for the full list of replication settings.
🎀 Options 🎀
--pgdata— Location of the Postgres node being managed locally. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl.--json— Output JSON formatted data.--formation— Show replication settings for given formation. Defaults todefault.--name— Show replication settings for given node, selected by name.
🌱 Environment 🌱
PGDATA— Postgres directory location. Can be used instead of the--pgdataoption.PG_AUTOCTL_MONITOR— Postgres URI to connect to the monitor node, can be used instead of the--monitoroption.XDG_CONFIG_HOME— Thepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification.XDG_DATA_HOME— Thepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification.
🌈 Examples 🌈
$ pg_autoctl set node candidate-priority --name node1 65
12:47:59 92326 INFO Waiting for the settings to have been applied to the monitor and primary node
12:47:59 92326 INFO New state is reported by node 1 "node1" (localhost:5501): "apply_settings"
12:47:59 92326 INFO Setting goal state of node 1 "node1" (localhost:5501) to primary after it applied replication properties change.
12:47:59 92326 INFO New state is reported by node 1 "node1" (localhost:5501): "primary"
65
$ pg_autoctl set node candidate-priority --name node1 50 --json
12:48:05 92450 INFO Waiting for the settings to have been applied to the monitor and primary node
12:48:05 92450 INFO New state is reported by node 1 "node1" (localhost:5501): "apply_settings"
12:48:05 92450 INFO Setting goal state of node 1 "node1" (localhost:5501) to primary after it applied replication properties change.
12:48:05 92450 INFO New state is reported by node 1 "node1" (localhost:5501): "primary"
{
"candidate-priority": 50
}🔥 pg_autoctl perform
pg_autoctl perform — Perform an action orchestrated by the monitor. main character energy tbh 👑
💖 pg_autoctl perform failover
pg_autoctl perform failover — Perform a failover for given formation and group.
✨ Synopsis ✨
This command starts a Postgres failover orchestration from the pg_auto_failover monitor:
usage: pg_autoctl perform failover [ --pgdata --formation --group ]
--pgdata path to data directory
--formation formation to target, defaults to 'default'
--group group to target, defaults to 0
--wait how many seconds to wait, default to 60💜 Description 💜
The pg_auto_failover monitor can be used to orchestrate a manual failover, sometimes also known as a switchover. When doing so, split-brain are prevented thanks to intermediary states being used in the Finite State Machine. 😌 (not us being safe about it, we love to see it)
The pg_autoctl perform failover command waits until the failover is known complete on the monitor, or until the hard-coded 60s timeout has passed.
Fr fr bestie, listen: the failover orchestration is done in the background by the monitor, so even if the pg_autoctl perform failover stops on the timeout, the failover orchestration continues at the monitor. It does not stop. It is not waiting on you. 🫶
🎀 Options 🎀
--pgdata— Location of the Postgres node being managed locally. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl.--formation— Formation to target for the operation. Defaults todefault.--group— Postgres group to target for the operation. Defaults to0, only Citus formations may have more than one group.--wait— How many seconds to wait for notifications about the promotion. The command stops when the promotion is finished (a node is primary), or when the timeout has elapsed, whichever comes first. The value0(zero) disables the timeout and allows the command to wait forever. 💀 (yes, literally forever — it's giving commitment issues in reverse)
🌍 Environment 🌍
PGDATA— where the Postgres directory lives. You can use this instead of the--pgdataoption, bestie, they're literally the same thing.PG_AUTOCTL_MONITOR— the Postgres URI to connect to the monitor node. Use it instead of the--monitoroption 💅PGHOST,PGPORT,PGDATABASE,PGUSER,PGCONNECT_TIMEOUT, ... — go read the Postgres docs about Environment Variables for the details, fr fr.TMPDIR— thepgcopydbcommand creates all its work files and directories in${TMPDIR}/pgcopydb, and it defaults to/tmp/pgcopydb.XDG_CONFIG_HOME— thepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification ✨XDG_DATA_HOME— thepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification ✨
🎀 Examples 🎀
$ pg_autoctl perform failover
12:57:30 3635 INFO Listening monitor notifications about state changes in formation "default" and group 0
12:57:30 3635 INFO Following table displays times when notifications are received
Time | Name | Node | Host:Port | Current State | Assigned State
---------+-------+-------+----------------+---------------------+--------------------
12:57:30 | node1 | 1 | localhost:5501 | primary | draining
12:57:30 | node1 | 1 | localhost:5501 | draining | draining
12:57:30 | node2 | 2 | localhost:5502 | secondary | report_lsn
12:57:30 | node3 | 3 | localhost:5503 | secondary | report_lsn
12:57:36 | node3 | 3 | localhost:5503 | report_lsn | report_lsn
12:57:36 | node2 | 2 | localhost:5502 | report_lsn | report_lsn
12:57:36 | node2 | 2 | localhost:5502 | report_lsn | prepare_promotion
12:57:36 | node2 | 2 | localhost:5502 | prepare_promotion | prepare_promotion
12:57:36 | node2 | 2 | localhost:5502 | prepare_promotion | stop_replication
12:57:36 | node1 | 1 | localhost:5501 | draining | demote_timeout
12:57:36 | node3 | 3 | localhost:5503 | report_lsn | join_secondary
12:57:36 | node1 | 1 | localhost:5501 | demote_timeout | demote_timeout
12:57:36 | node3 | 3 | localhost:5503 | join_secondary | join_secondary
12:57:37 | node2 | 2 | localhost:5502 | stop_replication | stop_replication
12:57:37 | node2 | 2 | localhost:5502 | stop_replication | wait_primary
12:57:37 | node1 | 1 | localhost:5501 | demote_timeout | demoted
12:57:37 | node1 | 1 | localhost:5501 | demoted | demoted
12:57:37 | node2 | 2 | localhost:5502 | wait_primary | wait_primary
12:57:37 | node3 | 3 | localhost:5503 | join_secondary | secondary
12:57:37 | node1 | 1 | localhost:5501 | demoted | catchingup
12:57:38 | node3 | 3 | localhost:5503 | secondary | secondary
12:57:38 | node2 | 2 | localhost:5502 | wait_primary | primary
12:57:38 | node1 | 1 | localhost:5501 | catchingup | catchingup
12:57:38 | node2 | 2 | localhost:5502 | primary | primary$ pg_autoctl show state
Name | Node | Host:Port | LSN | Connection | Current State | Assigned State
------+-------+----------------+-----------+--------------+---------------------+--------------------
node1 | 1 | localhost:5501 | 0/4000F50 | read-only | secondary | secondary
node2 | 2 | localhost:5502 | 0/4000F50 | read-write | primary | primary
node3 | 3 | localhost:5503 | 0/4000F50 | read-only | secondary | secondary👑 pg_autoctl perform switchover 👑
pg_autoctl perform switchover — do a switchover for the given formation and group. That's the whole vibe.
📖 Synopsis 📖
This command kicks off a Postgres switchover orchestration from the pg_auto_switchover monitor:
usage: pg_autoctl perform switchover [ --pgdata --formation --group ]
--pgdata path to data directory
--formation formation to target, defaults to 'default'
--group group to target, defaults to 0💖 Description 💖
So the pg_auto_switchover monitor can be used to orchestrate a manual switchover, sometimes also known as a switchover (yes bestie, same word twice, that's just how it is 😭). When it does that, split-brain gets prevented thanks to intermediary states being used in the Finite State Machine. No chaos allowed in this household 💅
The pg_autoctl perform switchover command waits until the switchover is known complete on the monitor, or until the hard-coded 60s timeout has passed. Whichever comes first, periodt.
The switchover orchestration is done in the background by the monitor, so even if the pg_autoctl perform switchover stops on the timeout, the switchover orchestration continues at the monitor. The monitor is not gonna stop just because your little command gave up 🙈
See also pg_autoctl perform failover, a synonym for this command.
✨ Options ✨
--pgdata— Location of the Postgres node being managed locally. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl.--formation— Formation to target for the operation. Defaults todefault.--group— Postgres group to target for the operation. Defaults to0, only Citus formations may have more than one group.
🌍 Environment 🌍
PGDATA— Postgres directory location. Can be used instead of the--pgdataoption.PG_AUTOCTL_MONITOR— Postgres URI to connect to the monitor node, can be used instead of the--monitoroption.PG_CONFIG— Can be set to the absolute path to thepg_configPostgres tool. This is mostly used in the context of building extensions, though it can be a useful way to select a Postgres version when several are installed on the same system.PATH— Used the usual way mostly. Some entries that are searched in thePATHby thepg_autoctlcommand are expected to be found only once, to avoid mistakes with Postgres major versions. 👀PGHOST,PGPORT,PGDATABASE,PGUSER,PGCONNECT_TIMEOUT, ... — See the Postgres docs about Environment Variables for details.TMPDIR— Thepgcopydbcommand creates all its work files and directories in${TMPDIR}/pgcopydb, and defaults to/tmp/pgcopydb.XDG_CONFIG_HOME— Thepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification.XDG_DATA_HOME— Thepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification.
🎀 Examples 🎀
$ pg_autoctl show state
Name | Node | Host:Port | LSN | Connection | Current State | Assigned State
------+-------+----------------+-----------+--------------+---------------------+--------------------
node1 | 1 | localhost:5501 | 0/4000F88 | read-only | secondary | secondary
node2 | 2 | localhost:5502 | 0/4000F88 | read-write | primary | primary
node3 | 3 | localhost:5503 | 0/4000F88 | read-only | secondary | secondary🌈 pg_autoctl perform promotion 🌈
pg_autoctl perform promotion — perform a failover that promotes a target node. Giving one specific node main character energy 👑
📖 Synopsis 📖
This command starts a Postgres failover orchestration from the pg_auto_promotion monitor and targets the given node:
usage: pg_autoctl perform promotion [ --pgdata --formation --group ]
--pgdata path to data directory
--formation formation to target, defaults to 'default'
--name node name to target, defaults to current node
--wait how many seconds to wait, default to 60💖 Description 💖
The pg_auto_promotion monitor can be used to orchestrate a manual promotion, sometimes also known as a switchover. And when it does that, split-brain gets prevented thanks to intermediary states being used in the Finite State Machine. Safety first, always 🫶
The pg_autoctl perform promotion command waits until the promotion is known complete on the monitor, or until the hard-coded 60s timeout has passed.
The promotion orchestration is done in the background by the monitor, so even if the pg_autoctl perform promotion stops on the timeout, the promotion orchestration continues at the monitor. It's giving "I'll finish this with or without you" 😭
✨ Options ✨
--pgdata— Location of the Postgres node being managed locally. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl.--formation— Formation to target for the operation. Defaults todefault.--name— Name of the node that should be elected as the new primary node.--wait— How many seconds to wait for notifications about the promotion. The command stops when the promotion is finished (a node is primary), or when the timeout has elapsed, whichever comes first. The value0(zero) disables the timeout and allows the command to wait forever. Bestie, forever is a long time, use that one with intent 💀
🌍 Environment 🌍
PGDATA— Postgres directory location. Can be used instead of the--pgdataoption.PG_AUTOCTL_MONITOR— Postgres URI to connect to the monitor node, can be used instead of the--monitoroption.PG_CONFIG— Can be set to the absolute path to thepg_configPostgres tool. This is mostly used in the context of building extensions, though it can be a useful way to select a Postgres version when several are installed on the same system.PATH— Used the usual way mostly. Some entries that are searched in thePATHby thepg_autoctlcommand are expected to be found only once, to avoid mistakes with Postgres major versions. 👀PGHOST,PGPORT,PGDATABASE,PGUSER,PGCONNECT_TIMEOUT, ... — See the Postgres docs about Environment Variables for details.TMPDIR— Thepgcopydbcommand creates all its work files and directories in${TMPDIR}/pgcopydb, and defaults to/tmp/pgcopydb.XDG_CONFIG_HOME— Thepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification.XDG_DATA_HOME— Thepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification.
$ pg_autoctl perform promotion --name node1
13:08:13 15297 INFO Listening monitor notifications about state changes in formation "default" and group 0
13:08:13 15297 INFO Following table displays times when notifications are received
Time | Name | Node | Host:Port | Current State | Assigned State
---------+-------+-------+----------------+---------------------+--------------------
13:08:13 | node1 | 0/1 | localhost:5501 | secondary | secondary
13:08:13 | node2 | 0/2 | localhost:5502 | primary | draining
13:08:13 | node2 | 0/2 | localhost:5502 | draining | draining
13:08:13 | node1 | 0/1 | localhost:5501 | secondary | report_lsn
13:08:13 | node3 | 0/3 | localhost:5503 | secondary | report_lsn
13:08:19 | node3 | 0/3 | localhost:5503 | report_lsn | report_lsn
13:08:19 | node1 | 0/1 | localhost:5501 | report_lsn | report_lsn
13:08:19 | node1 | 0/1 | localhost:5501 | report_lsn | prepare_promotion
13:08:19 | node1 | 0/1 | localhost:5501 | prepare_promotion | prepare_promotion
13:08:19 | node1 | 0/1 | localhost:5501 | prepare_promotion | stop_replication
13:08:19 | node2 | 0/2 | localhost:5502 | draining | demote_timeout
13:08:19 | node3 | 0/3 | localhost:5503 | report_lsn | join_secondary
13:08:19 | node2 | 0/2 | localhost:5502 | demote_timeout | demote_timeout
13:08:19 | node3 | 0/3 | localhost:5503 | join_secondary | join_secondary
13:08:20 | node1 | 0/1 | localhost:5501 | stop_replication | stop_replication
13:08:20 | node1 | 0/1 | localhost:5501 | stop_replication | wait_primary
13:08:20 | node2 | 0/2 | localhost:5502 | demote_timeout | demoted
13:08:20 | node1 | 0/1 | localhost:5501 | wait_primary | wait_primary
13:08:20 | node3 | 0/3 | localhost:5503 | join_secondary | secondary
13:08:20 | node2 | 0/2 | localhost:5502 | demoted | demoted
13:08:20 | node2 | 0/2 | localhost:5502 | demoted | catchingup
13:08:21 | node3 | 0/3 | localhost:5503 | secondary | secondary
13:08:21 | node1 | 0/1 | localhost:5501 | wait_primary | primary
13:08:21 | node2 | 0/2 | localhost:5502 | catchingup | catchingup
13:08:21 | node1 | 0/1 | localhost:5501 | primary | primary
$ pg_autoctl show state
Name | Node | Host:Port | LSN | Connection | Current State | Assigned State
------+-------+----------------+-----------+--------------+---------------------+--------------------
node1 | 1 | localhost:5501 | 0/40012F0 | read-write | primary | primary
node2 | 2 | localhost:5502 | 0/40012F0 | read-only | secondary | secondary
node3 | 3 | localhost:5503 | 0/40012F0 | read-only | secondary | secondaryOKAYYY bestie, look at that glow-up 💅✨ node1 starts out a humble secondary and ends the show as a full-on primary, node2 gets gracefully draining → demote_timeout → demoted → catchingup, and node3 just quietly re-join_secondarys like the supportive icon it is 🫶. Every single state transition is timestamped so you can literally watch the whole story unfold.
And then pg_autoctl show state does the vibe check and confirms it: node1 is read-write and primary (main character energy 👑), while node2 and node3 are both read-only secondary. Everybody on the same LSN 0/40012F0? Obsessed. That's a healthy formation fr fr 💖
pg_autoctl do 🔧
pg_autoctl do — Internal commands and internal QA tooling 🧪
Okay so the debug commands for pg_autoctl are only available when the environment variable PG_AUTOCTL_DEBUG is set (to any value). Any value! Even 1, even yes, even potato 🌸. Just set it and the secret menu opens.
When testing pg_auto_failover, it is helpful to be able to play with the local nodes using the same lower-level API as used by the pg_auto_failover Finite State Machine transitions 🔄. Some commands could be useful in contexts other than pg_auto_failover development and QA work, so some documentation has been made available 📚✨.
pg_autoctl do tmux 🖥️
pg_autoctl do tmux — Set of facilities to handle tmux interactive sessions 🌈
Synopsis 📋
pg_autoctl do tmux provides the following commands:
pg_autoctl do tmux
+ compose Set of facilities to handle docker compose sessions
script Produce a tmux script for a demo or a test case (debug only)
session Run a tmux session for a demo or a test case
stop Stop pg_autoctl processes that belong to a tmux session
wait Wait until a given node has been registered on the monitor
clean Clean-up a tmux session processes and root dir
pg_autoctl do tmux compose
config Produce a docker compose configuration file for a demo
script Produce a tmux script for a demo or a test case (debug only)
session Run a tmux session for a demo or a test caseDescription 💬
An easy way to get started with pg_auto_failover in a localhost only formation with three nodes is to run the following command ✨:
$ PG_AUTOCTL_DEBUG=1 pg_autoctl do tmux session \
--root /tmp/pgaf \
--first-pgport 9000 \
--nodes 4 \
--layout tiledThis requires the command tmux to be available in your PATH 🫶. The pg_autoctl do tmux session commands prepares a self-contained root directory where to create pg_auto_failover nodes and their configuration, then prepares a tmux script, and then runs the script with a command such as:
/usr/local/bin/tmux -v start-server ; source-file /tmp/pgaf/script-9000.tmuxThe tmux session contains a single tmux window multiple panes 🪟:
- one pane for the monitor 👀
- one pane per Postgres nodes, here 4 of them 🐘
- one pane for running
pg_autoctl watch📺 - one extra pane for an interactive shell 💻
Usually the first two commands to run in the interactive shell, once the formation is stable (one node is primary, the other ones are all secondary), are the following 🌟:
$ pg_autoctl get formation settings
$ pg_autoctl perform failoverUsing docker compose to run a distributed system 🐳
The same idea can also be implemented with docker compose to run the nodes, and still using tmux to have three control panels this time 🎛️:
- one pane for the
docker composecumulative logs of all the nodes 📜 - one pane for running
pg_autoctl watch📺 - one extra pane for an interactive shell 💻
For this setup, you can use the following command:
PG_AUTOCTL_DEBUG=1 pg_autoctl do tmux compose session \
--root ./tmux/citus \
--first-pgport 5600 \
--nodes 3 \
--async-nodes 0 \
--node-priorities 50,50,0 \
--sync-standbys -1 \
--citus-workers 4 \
--citus-secondaries 0 \
--citus \
--layout even-verticalThe pg_autoctl do tmux compose session command also takes care of creating external docker volumes and referencing them for each node in the docker compose file 💾✨
pg_autoctl do tmux session 🎬
This command runs a tmux session for a demo or a test case 🌸
usage: pg_autoctl do tmux session [option ...]
--root path where to create a cluster
--first-pgport first Postgres port to use (5500)
--nodes number of Postgres nodes to create (2)
--async-nodes number of async nodes within nodes (0)
--node-priorities list of nodes priorities (50)
--sync-standbys number-sync-standbys to set (0 or 1)
--skip-pg-hba use --skip-pg-hba when creating nodes
--citus start a Citus formation
--citus-workers number of Citus workers to create (2)
--citus-secondaries number of Citus secondaries to create (0)
--layout tmux layout to use (even-vertical)
--binpath path to the pg_autoctl binary (current binary path)pg_autoctl do tmux compose session 🐳🎬
This command runs a tmux session for a demo or a test case. It generates a docker compose file and then uses docker compose to drive many nodes 🚀
usage: pg_autoctl do tmux compose session [option ...]
--root path where to create a cluster
--first-pgport first Postgres port to use (5500)
--nodes number of Postgres nodes to create (2)
--async-nodes number of async nodes within nodes (0)
--node-priorities list of nodes priorities (50)
--sync-standbys number-sync-standbys to set (0 or 1)
--skip-pg-hba use --skip-pg-hba when creating nodes
--layout tmux layout to use (even-vertical)
--binpath path to the pg_autoctl binary (current binary path)pg_autoctl do demo 🎪
pg_autoctl do demo — Use a demo application for pg_auto_failover 🌟
Synopsis 📋
pg_autoctl do demo provides the following commands:
pg_autoctl do demo
run Run the pg_auto_failover demo application
uri Grab the application connection string from the monitor
ping Attempt to connect to the application URI
summary Display a summary of the previous demo app runTo run a demo, use pg_autoctl do demo run:
usage: pg_autoctl do demo run [option ...]
--monitor Postgres URI of the pg_auto_failover monitor
--formation Formation to use (default)
--group Group Id to failover (0)
--username PostgreSQL's username
--clients How many client processes to use (1)
--duration Duration of the demo app, in seconds (30)
--first-failover Timing of the first failover (10)
--failover-freq Seconds between subsequent failovers (45)Description 💬
The pg_autoctl debug tooling includes a demo application 🎉
The demo prepare its Postgres schema on the target database, and then starts several clients (see --clients) that concurrently connect to the target application URI and record the time it took to establish the Postgres connection to the current read-write node, with information about the retry policy metrics 📈✨
Example 🌰
$ pg_autoctl do demo run --monitor 'postgres://autoctl_node@localhost:5500/pg_auto_failover?sslmode=prefer' --clients 10
14:43:35 19660 INFO Using application connection string "postgres://localhost:5502,localhost:5503,localhost:5501/demo?target_session_attrs=read-write&sslmode=prefer"
14:43:35 19660 INFO Using Postgres user PGUSER "dim"
14:43:35 19660 INFO Preparing demo schema: drop schema if exists demo cascade
14:43:35 19660 WARN NOTICE: schema "demo" does not exist, skipping
14:43:35 19660 INFO Preparing demo schema: create schema demo
14:43:35 19660 INFO Preparing demo schema: create table demo.tracking(ts timestamptz default now(), client integer, loop integer, retries integer, us bigint, recovery bool)
14:43:36 19660 INFO Preparing demo schema: create table demo.client(client integer, pid integer, retry_sleep_ms integer, retry_cap_ms integer, failover_count integer)
14:43:36 19660 INFO Starting 10 concurrent clients as sub-processes
14:43:36 19675 INFO Failover client is started, will failover in 10s and every 45s after that
...Like, the demo literally drops and recreates its schema, spins up 10 concurrent clients as sub-processes, and then a failover client starts and will failover in 10s and every 45s after that 💀🔥 It's chaos but it's controlled chaos, iykyk.
$ pg_autoctl do demo summary --monitor 'postgres://autoctl_node@localhost:5500/pg_auto_failover?sslmode=prefer' --clients 10
14:44:27 22789 INFO Using application connection string "postgres://localhost:5503,localhost:5501,localhost:5502/demo?target_session_attrs=read-write&sslmode=prefer"
14:44:27 22789 INFO Using Postgres user PGUSER "dim"
14:44:27 22789 INFO Summary for the demo app running with 10 clients for 30s
Client | Connections | Retries | Min Connect Time (ms) | max | p95 | p99
----------------------+-------------+---------+-----------------------+----------+---------+---------
Client 1 | 136 | 14 | 58.318 | 2601.165 | 244.443 | 261.809
Client 2 | 136 | 5 | 55.199 | 2514.968 | 242.362 | 259.282
Client 3 | 134 | 6 | 55.815 | 2974.247 | 241.740 | 262.908
Client 4 | 135 | 7 | 56.542 | 2970.922 | 238.995 | 251.177
Client 5 | 136 | 8 | 58.339 | 2758.106 | 238.720 | 252.439
Client 6 | 134 | 9 | 58.679 | 2813.653 | 244.696 | 254.674
Client 7 | 134 | 11 | 58.737 | 2795.974 | 243.202 | 253.745
Client 8 | 136 | 12 | 52.109 | 2354.952 | 242.664 | 254.233
Client 9 | 137 | 19 | 59.735 | 2628.496 | 235.668 | 253.582
Client 10 | 133 | 6 | 57.994 | 3060.489 | 242.156 | 256.085
All Clients Combined | 1351 | 97 | 52.109 | 3060.489 | 241.848 | 258.450
(11 rows)
Min Connect Time (ms) | max | freq | bar
-----------------------+----------+------+-----------------------------------------------
52.109 | 219.105 | 1093 | ▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒▒
219.515 | 267.168 | 248 | ▒▒▒▒▒▒▒▒▒▒
2354.952 | 2354.952 | 1 |
2514.968 | 2514.968 | 1 |
2601.165 | 2628.496 | 2 |
2758.106 | 2813.653 | 3 |
2970.922 | 2974.247 | 2 |
3060.489 | 3060.489 | 1 |
(8 rows)pg_autoctl do service restart 🔄
pg_autoctl do service restart — run the pg_autoctl sub-processes (aka services) 💫
Synopsis
pg_autoctl do service restart hands you these commands, bestie:
pg_autoctl do service restart
postgres Restart the pg_autoctl postgres controller service
listener Restart the pg_autoctl monitor listener service
node-active Restart the pg_autoctl keeper node-active serviceDescription
Okay so here's the vibe: you can restart the pg_autoctl service or the Postgres service without messing with the other one that's still running. Iconic, honestly. 🫶
So like, if you just wanna bounce the pg_autoctl parts and leave Postgres completely unbothered:
$ pg_autoctl do service restart node-active --pgdata node1
14:52:06 31223 INFO Sending the TERM signal to service "node-active" with pid 26626
14:52:06 31223 INFO Service "node-active" has been restarted with pid 31230
31230And the Postgres service? Not. Touched. ✨ It was never impacted by the restart of the pg_autoctl process. Slay.
pg_autoctl do show 👀
pg_autoctl do show — show some debug level information 🕵️♀️
Synopsis
So both pg_autoctl create monitor and pg_autoctl create postgres do this lil automatic detection thing for the node network settings when you don't pass the option --hostname. Cute, right?
And adding onto that: when a new node gets registered to the monitor, the other nodes also edit their Postgres HBA rules so the new node can connect — unless you used the option --skip-pg-hba.
So basically the debug sub-commands under pg_autoctl do show are here so you can peek in detail at the network discovery that pg_autoctl is doing. 🔍
pg_autoctl do show gives you these commands:
pg_autoctl do show
ipaddr Print this node's IP address information
cidr Print this node's CIDR information
lookup Print this node's DNS lookup information
hostname Print this node's default hostname
reverse Lookup given hostname and check reverse DNS setuppg_autoctl do show ipaddr 🌐
Connects to an external IP address and uses getsockname(2) to grab the current address the socket is bound to.
The external IP address defaults to 8.8.8.8 — that's a Google-provided public DNS server, no cap — or it falls back to the monitor IP address / hostname when you're in the context of pg_autoctl create postgres.
$ pg_autoctl do show ipaddr
16:42:40 62631 INFO ipaddr.c:107: Connecting to 8.8.8.8 (port 53)
192.168.1.156pg_autoctl do show cidr 🧮
Connects to an external IP address the exact same way as the previous command pg_autoctl do show ipaddr does, and then matches the local socket name against the list of local network interfaces. When it finds a match, it uses the netmask of that interface to compute the CIDR notation from the IP address. Math girl summer. 💅
And that computed CIDR notation? It gets used in HBA rules.
$ pg_autoctl do show cidr
16:43:19 63319 INFO Connecting to 8.8.8.8 (port 53)
192.168.1.0/24pg_autoctl do show hostname 🏷️
Uses either its first (and only) argument, or the result of gethostname(2), as the candidate hostname for HBA rules — and then checks that the hostname actually resolves to an IP address belonging to one of the machine's network interfaces.
When the hostname forward-dns lookup resolves to an IP that's local to the node you're running the command on, then it does a reverse-lookup from that IP address to see if it matches the candidate hostname. 🦋
$ pg_autoctl do show hostname
DESKTOP-IC01GOOS.europe.corp.microsoft.com
$ pg_autoctl -vv do show hostname 'postgres://autoctl_node@localhost:5500/pg_auto_failover'
13:45:00 93122 INFO cli_do_show.c:256: Using monitor hostname "localhost" and port 5500
13:45:00 93122 INFO ipaddr.c:107: Connecting to ::1 (port 5500)
13:45:00 93122 DEBUG cli_do_show.c:272: cli_show_hostname: ip ::1
13:45:00 93122 DEBUG cli_do_show.c:283: cli_show_hostname: host localhost
13:45:00 93122 DEBUG cli_do_show.c:294: cli_show_hostname: ip ::1
localhostpg_autoctl do show lookup 🔎
Checks that the argument you gave it is a hostname that resolves to a local IP address — meaning an IP address that's associated with a local network interface.
$ pg_autoctl do show lookup DESKTOP-IC01GOOS.europe.corp.microsoft.com
DESKTOP-IC01GOOS.europe.corp.microsoft.com: 192.168.1.156pg_autoctl do show reverse 🔁
Implements the exact same DNS checks as the Postgres HBA matching code does: first a forward DNS lookup of the given hostname, and then a reverse-lookup from all the IP addresses it got back. It's a success when at least one of the IP addresses from the forward lookup resolves back to the given hostname (as the first answer to the reverse DNS lookup). 🌈
$ pg_autoctl do show reverse DESKTOP-IC01GOOS.europe.corp.microsoft.com
16:44:49 64910 FATAL Failed to find an IP address for hostname "DESKTOP-IC01GOOS.europe.corp.microsoft.com" that matches hostname again in a reverse-DNS lookup.
16:44:49 64910 INFO Continuing with IP address "192.168.1.156"
$ pg_autoctl -vv do show reverse DESKTOP-IC01GOOS.europe.corp.microsoft.com
16:44:45 64832 DEBUG ipaddr.c:719: DESKTOP-IC01GOOS.europe.corp.microsoft.com has address 192.168.1.156
16:44:45 64832 DEBUG ipaddr.c:733: reverse lookup for "192.168.1.156" gives "desktop-ic01goos.europe.corp.microsoft.com" first
16:44:45 64832 DEBUG ipaddr.c:719: DESKTOP-IC01GOOS.europe.corp.microsoft.com has address 192.168.1.156
16:44:45 64832 DEBUG ipaddr.c:733: reverse lookup for "192.168.1.156" gives "desktop-ic01goos.europe.corp.microsoft.com" first
16:44:45 64832 DEBUG ipaddr.c:719: DESKTOP-IC01GOOS.europe.corp.microsoft.com has address 2a01:110:10:40c::2ad
16:44:45 64832 DEBUG ipaddr.c:728: Failed to resolve hostname from address "192.168.1.156": nodename nor servname provided, or not known
16:44:45 64832 DEBUG ipaddr.c:719: DESKTOP-IC01GOOS.europe.corp.microsoft.com has address 2a01:110:10:40c::2ad
16:44:45 64832 DEBUG ipaddr.c:728: Failed to resolve hostname from address "192.168.1.156": nodename nor servname provided, or not known
16:44:45 64832 DEBUG ipaddr.c:719: DESKTOP-IC01GOOS.europe.corp.microsoft.com has address 100.64.34.213
16:44:45 64832 DEBUG ipaddr.c:728: Failed to resolve hostname from address "192.168.1.156": nodename nor servname provided, or not known
16:44:45 64832 DEBUG ipaddr.c:719: DESKTOP-IC01GOOS.europe.corp.microsoft.com has address 100.64.34.213
16:44:45 64832 DEBUG ipaddr.c:728: Failed to resolve hostname from address "192.168.1.156": nodename nor servname provided, or not known
16:44:45 64832 FATAL cli_do_show.c:333: Failed to find an IP address for hostname "DESKTOP-IC01GOOS.europe.corp.microsoft.com" that matches hostname again in a reverse-DNS lookup.
16:44:45 64832 INFO cli_do_show.c:334: Continuing with IP address "192.168.1.156"pg_autoctl do pgsetup 🛠️
pg_autoctl do pgsetup — manage a local Postgres setup 🐘
Synopsis
The main pg_autoctl commands implement low-level management tooling for a local Postgres instance, right? And some of those low-level Postgres commands can straight up be used as their own tool in certain cases. Werk. ✨
pg_autoctl do pgsetup provides you these commands:
pg_autoctl do pgsetup
pg_ctl Find a non-ambiguous pg_ctl program and Postgres version
discover Discover local PostgreSQL instance, if any
ready Return true is the local Postgres server is ready
wait Wait until the local Postgres server is ready
logs Outputs the Postgres startup logs
tune Compute and log some Postgres tuning optionspg_autoctl do pgsetup pg_ctl 🎛️
In a similar way to what -a does, this command scans your PATH for pg_ctl commands. Then it runs the pg_ctl --version command and parses the output to figure out which version of Postgres is available in the path. Detective work but make it database. 💅
$ pg_autoctl do pgsetup pg_ctl --pgdata node1
16:49:18 69684 INFO Environment variable PG_CONFIG is set to "/Applications/Postgres.app//Contents/Versions/12/bin/pg_config"
16:49:18 69684 INFO `pg_autoctl create postgres` would use "/Applications/Postgres.app/Contents/Versions/12/bin/pg_ctl" for Postgres 12.3
16:49:18 69684 INFO `pg_autoctl create monitor` would use "/Applications/Postgres.app/Contents/Versions/12/bin/pg_ctl" for Postgres 12.3pg_autoctl do pgsetup discover 🔭
Give it a PGDATA or a --pgdata option, and the command discovers whether a running Postgres service matches the pg_autoctl setup — and prints out all the info that pg_autoctl typically needs when it's managing a Postgres instance. 🫶
okay so here's literally what that discovery thing spits out at you, bestie:
$ pg_autoctl do pgsetup discover --pgdata node1
pgdata: /Users/dim/dev/MS/pg_auto_failover/tmux/node1
pg_ctl: /Applications/Postgres.app/Contents/Versions/12/bin/pg_ctl
pg_version: 12.3
pghost: /tmp
pgport: 5501
proxyport: 0
pid: 21029
is in recovery: no
Control Version: 1201
Catalog Version: 201909212
System Identifier: 6942422768095393833
Latest checkpoint LSN: 0/4059C18
Postmaster status: readypg_autoctl do pgsetup ready
it's giving pg_isready energy ✨ except it grabs the Postgres specs straight out of your pg_autoctl node setup instead. lowkey iconic.
$ pg_autoctl do pgsetup ready --pgdata node1
16:50:08 70582 INFO Postgres status is: "ready"pg_autoctl do pgsetup wait
So when pg_autoctl do pgsetup ready would come back all false because Postgres isn't ready yet 😭, this one just keeps probing every single second for 30 seconds, and dips out the instant Postgres is ready. no cap.
$ pg_autoctl do pgsetup wait --pgdata node1
16:50:22 70829 INFO Postgres is now serving PGDATA "/Users/dim/dev/MS/pg_auto_failover/tmux/node1" on port 5501 with pid 21029
16:50:22 70829 INFO Postgres status is: "ready"pg_autoctl do pgsetup logs
Prints the Postgres logs from the freshest log file chillin' in the PGDATA/log directory 🍓. full dump, here we go:
$ pg_autoctl do pgsetup logs --pgdata node1
16:50:39 71126 WARN Postgres logs from "/Users/dim/dev/MS/pg_auto_failover/tmux/node1/startup.log":
16:50:39 71126 INFO 2021-03-22 14:43:48.911 CET [21029] LOG: starting PostgreSQL 12.3 on x86_64-apple-darwin16.7.0, compiled by Apple LLVM version 8.1.0 (clang-802.0.42), 64-bit
16:50:39 71126 INFO 2021-03-22 14:43:48.913 CET [21029] LOG: listening on IPv6 address "::", port 5501
16:50:39 71126 INFO 2021-03-22 14:43:48.913 CET [21029] LOG: listening on IPv4 address "0.0.0.0", port 5501
16:50:39 71126 INFO 2021-03-22 14:43:48.913 CET [21029] LOG: listening on Unix socket "/tmp/.s.PGSQL.5501"
16:50:39 71126 INFO 2021-03-22 14:43:48.931 CET [21029] LOG: redirecting log output to logging collector process
16:50:39 71126 INFO 2021-03-22 14:43:48.931 CET [21029] HINT: Future log output will appear in directory "log".
16:50:39 71126 WARN Postgres logs from "/Users/dim/dev/MS/pg_auto_failover/tmux/node1/log/postgresql-2021-03-22_144348.log":
16:50:39 71126 INFO 2021-03-22 14:43:48.937 CET [21033] LOG: database system was shut down at 2021-03-22 14:43:46 CET
16:50:39 71126 INFO 2021-03-22 14:43:48.937 CET [21033] LOG: entering standby mode
16:50:39 71126 INFO 2021-03-22 14:43:48.942 CET [21033] LOG: consistent recovery state reached at 0/4022E88
16:50:39 71126 INFO 2021-03-22 14:43:48.942 CET [21033] LOG: invalid record length at 0/4022E88: wanted 24, got 0
16:50:39 71126 INFO 2021-03-22 14:43:48.946 CET [21029] LOG: database system is ready to accept read only connections
16:50:39 71126 INFO 2021-03-22 14:43:49.032 CET [21038] LOG: fetching timeline history file for timeline 4 from primary server
16:50:39 71126 INFO 2021-03-22 14:43:49.037 CET [21038] LOG: started streaming WAL from primary at 0/4000000 on timeline 3
16:50:39 71126 INFO 2021-03-22 14:43:49.046 CET [21038] LOG: replication terminated by primary server
16:50:39 71126 INFO 2021-03-22 14:43:49.046 CET [21038] DETAIL: End of WAL reached on timeline 3 at 0/4022E88.
16:50:39 71126 INFO 2021-03-22 14:43:49.047 CET [21033] LOG: new target timeline is 4
16:50:39 71126 INFO 2021-03-22 14:43:49.049 CET [21038] LOG: restarted WAL streaming at 0/4000000 on timeline 4
16:50:39 71126 INFO 2021-03-22 14:43:49.210 CET [21033] LOG: redo starts at 0/4022E88
16:50:39 71126 INFO 2021-03-22 14:52:06.692 CET [21029] LOG: received SIGHUP, reloading configuration files
16:50:39 71126 INFO 2021-03-22 14:52:06.906 CET [21029] LOG: received SIGHUP, reloading configuration files
16:50:39 71126 FATAL 2021-03-22 15:34:24.920 CET [21038] FATAL: terminating walreceiver due to timeout
16:50:39 71126 INFO 2021-03-22 15:34:24.973 CET [21033] LOG: invalid record length at 0/4059CC8: wanted 24, got 0
16:50:39 71126 INFO 2021-03-22 15:34:25.105 CET [35801] LOG: started streaming WAL from primary at 0/4000000 on timeline 4
16:50:39 71126 FATAL 2021-03-22 16:12:56.918 CET [35801] FATAL: terminating walreceiver due to timeout
16:50:39 71126 INFO 2021-03-22 16:12:57.086 CET [38741] LOG: started streaming WAL from primary at 0/4000000 on timeline 4
16:50:39 71126 FATAL 2021-03-22 16:23:39.349 CET [38741] FATAL: terminating walreceiver due to timeout
16:50:39 71126 INFO 2021-03-22 16:23:39.497 CET [41635] LOG: started streaming WAL from primary at 0/4000000 on timeline 4pg_autoctl do pgsetup tune
Spits out the pg_autoctl automated tuning options 💅. Depending on how many CPUs and how much RAM it sniffs out in whatever environment you're running it in, pg_autoctl can nudge some very basic Postgres tuning knobs so you've got a starting point.
$ pg_autoctl do pgsetup tune --pgdata node1 -vv
13:25:25 77185 DEBUG pgtuning.c:85: Detected 12 CPUs and 16 GB total RAM on this server
13:25:25 77185 DEBUG pgtuning.c:225: Setting autovacuum_max_workers to 3
13:25:25 77185 DEBUG pgtuning.c:228: Setting shared_buffers to 4096 MB
13:25:25 77185 DEBUG pgtuning.c:231: Setting work_mem to 24 MB
13:25:25 77185 DEBUG pgtuning.c:235: Setting maintenance_work_mem to 512 MB
13:25:25 77185 DEBUG pgtuning.c:239: Setting effective_cache_size to 12 GB
# basic tuning computed by pg_auto_failover
track_functions = pl
shared_buffers = '4096 MB'
work_mem = '24 MB'
maintenance_work_mem = '512 MB'
effective_cache_size = '12 GB'
autovacuum_max_workers = 3
autovacuum_vacuum_scale_factor = 0.08
autovacuum_analyze_scale_factor = 0.02okay now the low-level API is made available through the following pg_autoctl do commands, and — heads up bestie — these are only available in debug environments 🙈:
pg_autoctl do
+ monitor Query a pg_auto_failover monitor
+ fsm Manually manage the keeper's state
+ primary Manage a PostgreSQL primary server
+ standby Manage a PostgreSQL standby server
+ show Show some debug level information
+ pgsetup Manage a local Postgres setup
+ pgctl Signal the pg_autoctl postgres service
+ service Run pg_autoctl sub-processes (services)
+ tmux Set of facilities to handle tmux interactive sessions
+ azure Manage a set of Azure resources for a pg_auto_failover demo
+ demo Use a demo application for pg_auto_failoverand the rest of the whole menu, full send:
pg_autoctl do monitor
+ get Get information from the monitor
register Register the current node with the monitor
active Call in the pg_auto_failover Node Active protocol
version Check that monitor version is 1.5.0.1; alter extension update if not
parse-notification parse a raw notification message
pg_autoctl do monitor get
primary Get the primary node from pg_auto_failover in given formation/group
others Get the other nodes from the pg_auto_failover group of hostname/port
coordinator Get the coordinator node from the pg_auto_failover formation
pg_autoctl do fsm
init Initialize the keeper's state on-disk
state Read the keeper's state from disk and display it
list List reachable FSM states from current state
gv Output the FSM as a .gv program suitable for graphviz/dot
assign Assign a new goal state to the keeper
step Make a state transition if instructed by the monitor
+ nodes Manually manage the keeper's nodes list
pg_autoctl do fsm nodes
get Get the list of nodes from file (see --disable-monitor)
set Set the list of nodes to file (see --disable-monitor)
pg_autoctl do primary
+ slot Manage replication slot on the primary server
+ adduser Create users on primary
defaults Add default settings to postgresql.conf
identify Run the IDENTIFY_SYSTEM replication command on given host
pg_autoctl do primary slot
create Create a replication slot on the primary server
drop Drop a replication slot on the primary server
pg_autoctl do primary adduser
monitor add a local user for queries from the monitor
replica add a local user with replication privileges
pg_autoctl do standby
init Initialize the standby server using pg_basebackup
rewind Rewind a demoted primary server using pg_rewind
promote Promote a standby server to become writable
pg_autoctl do show
ipaddr Print this node's IP address information
cidr Print this node's CIDR information
lookup Print this node's DNS lookup information
hostname Print this node's default hostname
reverse Lookup given hostname and check reverse DNS setup
pg_autoctl do pgsetup
pg_ctl Find a non-ambiguous pg_ctl program and Postgres version
discover Discover local PostgreSQL instance, if any
ready Return true is the local Postgres server is ready
wait Wait until the local Postgres server is ready
logs Outputs the Postgres startup logs
tune Compute and log some Postgres tuning options
pg_autoctl do pgctl
on Signal pg_autoctl postgres service to ensure Postgres is running
off Signal pg_autoctl postgres service to ensure Postgres is stopped
pg_autoctl do service
+ getpid Get the pid of pg_autoctl sub-processes (services)
+ restart Restart pg_autoctl sub-processes (services)
pgcontroller pg_autoctl supervised postgres controller
postgres pg_autoctl service that start/stop postgres when asked
listener pg_autoctl service that listens to the monitor notifications
node-active pg_autoctl service that implements the node active protocol
pg_autoctl do service getpid
postgres Get the pid of the pg_autoctl postgres controller service
listener Get the pid of the pg_autoctl monitor listener service
node-active Get the pid of the pg_autoctl keeper node-active service
pg_autoctl do service restart
postgres Restart the pg_autoctl postgres controller service
listener Restart the pg_autoctl monitor listener service
node-active Restart the pg_autoctl keeper node-active service
pg_autoctl do tmux
script Produce a tmux script for a demo or a test case (debug only)
session Run a tmux session for a demo or a test case
stop Stop pg_autoctl processes that belong to a tmux session
wait Wait until a given node has been registered on the monitor
clean Clean-up a tmux session processes and root dirperiodt 💖 those are the debug-only goodies, bestie — handle with care 🫶
okay so the do azure family, let's go — these are the cloud-bestie commands 💙✨
pg_autoctl do azure
+ provision provision azure resources for a pg_auto_failover demo
+ tmux Run a tmux session with an Azure setup for QA/testing
+ show show azure resources for a pg_auto_failover demo
deploy Deploy a pg_autoctl VMs, given by name
create Create an azure QA environment
drop Drop an azure QA environment: resource group, network, VMs
ls List resources in a given azure region
ssh Runs ssh -l ha-admin <public ip address> for a given VM name
sync Rsync pg_auto_failover sources on all the target region VMs
pg_autoctl do azure provision
region Provision an azure region: resource group, network, VMs
nodes Provision our pre-created VM with pg_autoctl Postgres nodes
pg_autoctl do azure tmux
session Create or attach a tmux session for the created Azure VMs
kill Kill an existing tmux session for Azure VMs
pg_autoctl do azure show
ips Show public and private IP addresses for selected VMs
state Connect to the monitor node to show the current state
pg_autoctl do demo
run Run the pg_auto_failover demo application
uri Grab the application connection string from the monitor
ping Attempt to connect to the application URI
summary Display a summary of the previous demo app runpg_autoctl run 🔥
pg_autoctl run — Run the pg_autoctl service (monitor or keeper)
✨ Synopsis ✨
okay so this command literally starts up the processes you need to run a monitor node or a keeper node, and which one you get depends on the configuration file that belongs to the --pgdata option or the PGDATA environment variable. slay.
usage: pg_autoctl run [ --pgdata --name --hostname --pgport ]
--pgdata path to data directory
--name pg_auto_failover node name
--hostname hostname used to connect from other nodes
--pgport PostgreSQL's port number💖 Description 💖
so when you register Postgres nodes to the pg_auto_failover monitor using the pg_autoctl create postgres command, the nodes get registered with metadata: the node name, hostname and Postgres port. that's the whole vibe.
the node name is used mostly in the logs and pg_autoctl show state commands, and it helps human administrators of the formation. think of it as the cute little name tag 🏷️ for your node.
the node hostname and pgport are used by other nodes, including the pg_auto_failover monitor, to open a Postgres connection.
both the node name and the node hostname and port can be changed after the node registration by using either this command (pg_autoctl run) or the pg_autoctl config set command. glow up your node whenever ✨
🎀 Options 🎀
--pgdata— Location of the Postgres node being managed locally. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl.--name— Node name used on the monitor to refer to this node. The hostname is a technical information, and given Postgres requirements on the HBA setup and DNS resolution (both forward and reverse lookups), IP addresses are often used for the hostname.
The --name option allows using a user-friendly name for your Postgres nodes. cute name, serious infra 💅
--hostname— Hostname or IP address (both v4 and v6 are supported) to use from any other node to connect to this node.
When not provided, a default value is computed by running the following algorithm.
- We get this machine's "public IP" by opening a connection to the given monitor hostname or IP address. Then we get TCP/IP client address that has been used to make that connection.
- We then do a reverse DNS lookup on the IP address found in the previous step to fetch a hostname for our local machine.
- If the reverse DNS lookup is successful , then
pg_autoctldoes a forward DNS lookup of that hostname.
When the forward DNS lookup response in step 3. is an IP address found in one of our local network interfaces, then pg_autoctl uses the hostname found in step 2. as the default --hostname. Otherwise it uses the IP address found in step 1.
You may use the --hostname command line option to bypass the whole DNS lookup based process and force the local node name to a fixed value. skip the drama, just set it 🦋
--pgport— Postgres port to use, defaults to5432.
pg_autoctl watch 👑
pg_autoctl watch — Display an auto-updating dashboard
✨ Synopsis ✨
This command outputs the events that the pg_auto_failover events records about state changes of the pg_auto_failover nodes managed by the monitor:
usage: pg_autoctl watch [ --pgdata --formation --group ]
--pgdata path to data directory
--monitor show the monitor uri
--formation formation to query, defaults to 'default'
--group group to query formation, defaults to all
--json output data in the JSON format🎀 Options 🎀
--pgdata— Location of the Postgres node being managed locally. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl.--monitor— Postgres URI used to connect to the monitor. Must use theautoctl_nodeusername and target thepg_auto_failoverdatabase name. It is possible to show the Postgres URI from the monitor node using the commandpg_autoctl show uri.--formation— List the events recorded for nodes in the given formation. Defaults todefault.--group— Limit output to a single group in the formation. Default to including all groups registered in the target formation.
🌸 Environment 🌸
PGDATA— Postgres directory location. Can be used instead of the--pgdataoption.PG_AUTOCTL_MONITOR— Postgres URI to connect to the monitor node, can be used instead of the--monitoroption.XDG_CONFIG_HOME— Thepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification.XDG_DATA_HOME— Thepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification.
💖 Description 💖
the pg_autoctl watch output is divided in 3 sections. iconic layout, no cap.
Section one is a single header line which includes the name of the currently selected formation, the formation replication setting Number Sync Standbys, and then in the right most position the current time. 🕐
Section two displays one line per node, and each line contains a list of columns that describe the current state for the node. This list can includes the following columns, and which columns are part of the output depends on the terminal window size. This choice is dynamic and changes if your terminal window size changes. resize the terminal, resize the vibe 🌈
- Name — Name of the node.
- Node, or Id — Node information. When the formation has a single group (group zero), then this column only contains the
nodeId. Only Citus formations allow several groups. When using a Citus formation the Node column contains thegroupIdand thenodeId, separated by a colon, such as0:1for the first coordinator node. - Last Report, or Report — Time interval between now and the last known time when a node has reported to the monitor, using the
node_activeprotocol. This value is expected to stay under2sor abouts, and is known to increment when either thepg_autoctl runservice is not running, or when there is a network split. - Last Check, or Check — Time interval between now and the last known time when the monitor could connect to a node's Postgres instance, via its health check mechanism. This value is known to increment when either the Postgres service is not running on the target node, when there is a network split, or when the internal machinery (the health check worker background process) implements jitter.
- Host:Port — Hostname and port number used to connect to the node.
- TLI: LSN — Timeline identifier (TLI) and Postgres Log Sequence Number (LSN). The LSN is the current position in the Postgres WAL stream. This is a hexadecimal number. See
pg_lsnfor more information. The current timeline is incremented each time a failover happens, or when doing Point In Time Recovery. A node can only reach the secondary state when it is on the same timeline as its primary node. - Connection — This output field contains two bits of information. First, the Postgres connection type that the node provides, either
read-writeorread-only. Then the mark!is added when the monitor has failed to connect to this node, and?when the monitor didn't connect to the node yet. - Reported State — The current FSM state as reported to the monitor by the
pg_autoctlprocess running on the Postgres node. - Assigned State — The assigned FSM state on the monitor. When the assigned state is not the same as the reported start, then the
pg_autoctlprocess running on the Postgres node might have not retrieved the assigned state yet, or might still be implementing the FSM transition from the current state to the assigned state.
Section three (the third and last one, bestie) lists the most recent events that the monitor has registered, the more recent event is found at the bottom of the screen.
To quit the command hit either theF1key or theqkey. 🫶
pg_autoctl stop 💅
pg_autoctl stop — signal the pg_autoctl service for it to stop
✨ Synopsis ✨
This commands stops the processes needed to run a monitor node or a keeper node, depending on the configuration file that belongs to the --pgdata option or PGDATA environment variable.
usage: pg_autoctl stop [ --pgdata --fast --immediate ]
--pgdata path to data directory
--fast fast shutdown mode for the keeper
--immediate immediate shutdown mode for the keeper💖 Description 💖
the pg_autoctl stop commands finds the PID of the running service for the given --pgdata, and if the process is still running, sends a SIGTERM signal to the process. bye bestie, we're logging off 👋✨
So like, when pg_autoclt catches a shutdown signal? bestie, it's over, a whole shutdown sequence gets triggered 💅 and depending on which signal it caught, anything already in flight (like, oh idk, a state transition, no big deal) gets one of three fates:
- it runs allll the way to completion ✨
- it stops at the very next opportunity, no drama
- it stops immediately, like mid-transition and everything 😭 no takebacks
✨ Options ✨
--pgdata— where the Postgres node you're managing locally lives. Defaults to the environment variablePGDATA. And listen: use--monitorif you wanna connect to a monitor from anywhere, instead of the monitor URI that a local Postgres node managed withpg_autoctlwould use. 🌍--fast— Fast Shutdown mode forpg_autoctl. It sends theSIGINTsignal to the running service, which is literally the same thing as hittingC-con an interactive process running as a foreground shell job. 🏃♀️--immediate— Immediate Shutdown mode forpg_autoctl. Sends theSIGQUITsignal to the running service. No waiting, no mercy 💀
🌸 Environment 🌸
PGDATA— Postgres directory location. You can use this instead of the--pgdataoption. 🫶PG_AUTOCTL_MONITOR— Postgres URI to connect to the monitor node, usable instead of the--monitoroption. 🔗XDG_CONFIG_HOME— thepg_autoctlcommand keeps its configuration files in the standard place,XDG_CONFIG_HOME. See the XDG Base Directory Specification. 📁XDG_DATA_HOME— thepg_autoctlcommand keeps its internal states files in the standard place,XDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification. 💾
🌈 pg_autoctl reload 🌈
pg_autoctl reload — signal the pg_autoctl for it to reload its configuration 🔄
📋 Synopsis 📋
This command signals a running pg_autoctl process to reload its configuration from disk, and also signals the managed Postgres service to reload its configuration. two for one, we love to see it ✨
usage: pg_autoctl reload [ --pgdata ] [ --json ]
--pgdata path to data directory💖 Description 💖
The pg_autoctl reload commands finds the PID of the running service for the given --pgdata, and if the process is still running, sends a SIGHUP signal to the process. gentle little nudge, she'll get the message 🎀
✨ Options ✨
--pgdata— where the Postgres node you're managing locally lives. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl. 🌍
🌟 pg_autoctl status 🌟
pg_autoctl status — Display the current status of the pg_autoctl service 👑
📋 Synopsis 📋
This command outputs the current process status for the pg_autoctl service running for the given --pgdata location.
usage: pg_autoctl status [ --pgdata ] [ --json ]
--pgdata path to data directory
--json output data in the JSON format✨ Options ✨
--pgdata— Location of the Postgres node being managed locally. Defaults to the environment variablePGDATA.--json— Output a JSON formatted data instead of a table formatted list. for the data girlies 🧚
🌸 Environment 🌸
PGDATA— Postgres directory location. Can be used instead of the--pgdataoption.PG_AUTOCTL_MONITOR— Postgres URI to connect to the monitor node.XDG_CONFIG_HOME— thepg_autoctlcommand stores its configuration files in the standard placeXDG_CONFIG_HOME. See the XDG Base Directory Specification.XDG_DATA_HOME— thepg_autoctlcommand stores its internal states files in the standard placeXDG_DATA_HOME, which defaults to~/.local/share. See the XDG Base Directory Specification.
🎀 Example 🎀
okay bestie, look how pretty this output is:
$ pg_autoctl status --pgdata node1
11:26:30 27248 INFO pg_autoctl is running with pid 26618
11:26:30 27248 INFO Postgres is serving PGDATA "/Users/dim/dev/MS/pg_auto_failover/tmux/node1" on port 5501 with pid 26725
$ pg_autoctl status --pgdata node1 --json
11:26:37 27385 INFO pg_autoctl is running with pid 26618
11:26:37 27385 INFO Postgres is serving PGDATA "/Users/dim/dev/MS/pg_auto_failover/tmux/node1" on port 5501 with pid 26725
{
"postgres": {
"pgdata": "\/Users\/dim\/dev\/MS\/pg_auto_failover\/tmux\/node1",
"pg_ctl": "\/Applications\/Postgres.app\/Contents\/Versions\/12\/bin\/pg_ctl",
"version": "12.3",
"host": "\/tmp",
"port": 5501,
"proxyport": 0,
"pid": 26725,
"in_recovery": false,
"control": {
"version": 0,
"catalog_version": 0,
"system_identifier": "0"
},
"postmaster": {
"status": "ready"
}
},
"pg_autoctl": {
"pid": 26618,
"status": "running",
"pgdata": "\/Users\/dim\/dev\/MS\/pg_auto_failover\/tmux\/node1",
"version": "1.5.0",
"semId": 196609,
"services": [
{
"name": "postgres",
"pid": 26625,
"status": "running",
"version": "1.5.0",
"pgautofailover": "1.5.0.1"
},
{
"name": "node-active",
"pid": 26626,
"status": "running",
"version": "1.5.0",
"pgautofailover": "1.5.0.1"
}
]
}
}💫 pg_autoctl activate 💫
pg_autoctl activate — Activate a Citus worker from the Citus coordinator 🐝
📋 Synopsis 📋
This command calls the Citus "activation" API so that a node can be used to host shards for your reference and distributed tables.
usage: pg_autoctl activate [ --pgdata ]
--pgdata path to data directory💖 Description 💖
When you're creating a Citus worker, pg_autoctl create worker automatically activates the worker node to the coordinator. so like, you only need this command when something unexpected have happened and you wanna manually make sure the worker node has been activated at the Citus coordinator level. 🙈
Starting with Citus 10 it's also possible to activate the coordinator itself as a node with shard placement 👑 Use pg_autoctl activate on your Citus coordinator node manually to use that feature.
heads up bestie: when the Citus coordinator is activated, an extra step is then needed for it to host shards of distributed tables. If you want your coordinator to have shards, then have a look at the Citus API —citus_set_node_property— to set theshouldhaveshardsproperty totrue. no cap, don't skip this one 🌈
✨ Options ✨
--pgdata— Location of the Postgres node being managed locally. Defaults to the environment variablePGDATA. Use--monitorto connect to a monitor from anywhere, rather than the monitor URI used by a local Postgres node managed withpg_autoctl. 🌍
💖 CONFIGURING PG_AUTO_FAILOVER 💖
okay bestie so like, pg_auto_failover ships with a bunch of default settings and you can absolutely review them and change them depending on what trade-offs you wanna make in your own production setup. lowkey important. the settings you can change will have an impact on these operations:
- Deciding when to promote the secondary
pg_auto_failover decides to implement a failover to the secondary node when it detects that the primary node is unhealthy. Changing the following settings will have an impact on when the pg_auto_failover monitor decides to promote the secondary PostgreSQL node:
pgautofailover.health_check_max_retries
pgautofailover.health_check_period
pgautofailover.health_check_retry_delay
pgautofailover.health_check_timeout
pgautofailover.node_considered_unhealthy_timeout- Time taken to promote the secondary
At secondary promotion time, pg_auto_failover waits for the following timeout to make sure that all pending writes on the primary server made it to the secondary at shutdown time, thus preventing data loss.:
pgautofailover.primary_demote_timeout- Preventing promotion of the secondary
pg_auto_failover implements a trade-off where data availability trumps service availability. When the primary node of a PostgreSQL service is detected unhealthy, the secondary is only promoted if it was known to be eligible at the moment when the primary is lost.
In the case when synchronous replication was in use at the moment when the primary node is lost, then we know we can switch to the secondary safely, and the wal lag is 0 in that case.
In the case when the secondary server had been detected unhealthy before, then the pg_auto_failover monitor switches it from the state SECONDARY to the state CATCHING-UP and promotion is prevented then.
The following setting allows to still promote the secondary, allowing for a window of data loss:
pgautofailover.promote_wal_log_threshold🧚 pg_auto_failover Monitor 🧚
The configuration for the behavior of the monitor happens in the PostgreSQL database where the extension has been deployed:
pg_auto_failover=> select name, setting, unit, short_desc from pg_settings where name ~ 'pgautofailover.';
-[ RECORD 1 ]----------------------------------------------------------------------------------------------------
name | pgautofailover.enable_sync_wal_log_threshold
setting | 16777216
unit |
short_desc | Don't enable synchronous replication until secondary xlog is within this many bytes of the primary's
-[ RECORD 2 ]----------------------------------------------------------------------------------------------------
name | pgautofailover.health_check_max_retries
setting | 2
unit |
short_desc | Maximum number of re-tries before marking a node as failed.
-[ RECORD 3 ]----------------------------------------------------------------------------------------------------
name | pgautofailover.health_check_period
setting | 5000
unit | ms
short_desc | Duration between each check (in milliseconds).
-[ RECORD 4 ]----------------------------------------------------------------------------------------------------
name | pgautofailover.health_check_retry_delay
setting | 2000
unit | ms
short_desc | Delay between consecutive retries.
-[ RECORD 5 ]----------------------------------------------------------------------------------------------------
name | pgautofailover.health_check_timeout
setting | 5000
unit | ms
short_desc | Connect timeout (in milliseconds).
-[ RECORD 6 ]----------------------------------------------------------------------------------------------------
name | pgautofailover.node_considered_unhealthy_timeout
setting | 20000
unit | ms
short_desc | Mark node unhealthy if last ping was over this long ago
-[ RECORD 7 ]----------------------------------------------------------------------------------------------------
name | pgautofailover.primary_demote_timeout
setting | 30000
unit | ms
short_desc | Give the primary this long to drain before promoting the secondary
-[ RECORD 8 ]----------------------------------------------------------------------------------------------------
name | pgautofailover.promote_wal_log_threshold
setting | 16777216
unit |
short_desc | Don't promote secondary unless xlog is with this many bytes of the master
-[ RECORD 9 ]----------------------------------------------------------------------------------------------------
name | pgautofailover.startup_grace_period
setting | 10000
unit | ms
short_desc | Wait for at least this much time after startup before initiating a failover.You can edit the parameters as usual with PostgreSQL, either in the postgresql.conf file or using ALTER DATABASE pg_auto_failover SET parameter = value; commands, then issuing a reload. so easy bestie 💅
🎀 pg_auto_failover Keeper Service 🎀
For an introduction to the pg_autoctl commands relevant to the pg_auto_failover Keeper configuration, please see pg_autoctl config.
An example configuration file looks like the following:
[pg_autoctl]
role = keeper
monitor = postgres://autoctl_node@192.168.1.34:6000/pg_auto_failover
formation = default
group = 0
hostname = node1.db
nodekind = standalone
[postgresql]
pgdata = /data/pgsql/
pg_ctl = /usr/pgsql-10/bin/pg_ctl
dbname = postgres
host = /tmp
port = 5000
[replication]
slot = pgautofailover_standby
maximum_backup_rate = 100M
backup_directory = /data/backup/node1.db
[timeout]
network_partition_timeout = 20
postgresql_restart_failure_timeout = 20
postgresql_restart_failure_max_retries = 3To output, edit and check entries of the configuration, the following commands are provided:
pg_autoctl config check [--pgdata <pgdata>]
pg_autoctl config get [--pgdata <pgdata>] section.option
pg_autoctl config set [--pgdata <pgdata>] section.option valueThe [postgresql] section is discovered automatically by the pg_autoctl command and is not intended to be changed manually. hands off that one, she runs herself ✨
pg_autoctl.monitor
PostgreSQL service URL of the pg_auto_failover monitor, as given in the output of the pg_autoctl show uri command.
pg_autoctl.formation
A single pg_auto_failover monitor may handle several postgres formations. The default formation name default is usually fine.
pg_autoctl.group
This information is retrieved by the pg_auto_failover keeper when registering a node to the monitor, and should not be changed afterwards. Use at your own risk.
⚠️ "use at your own risk" is doing a LOT of heavy lifting in that sentence bestie. just don't touch it.
pg_autoctl.hostname
Node hostname used by all the other nodes in the cluster to contact this node. In particular, if this node is a primary then its standby uses that address to setup streaming replication.
replication.slot
Name of the PostgreSQL replication slot used in the streaming replication setup automatically deployed by pg_auto_failover. Replication slots can't be renamed in PostgreSQL. fr fr, no takebacks 🙈
replication.maximum_backup_rate
When pg_auto_failover (re-)builds a standby node using the pg_basebackup command, this parameter is given to pg_basebackup to throttle the network bandwidth used. Defaults to 100Mbps.
replication.backup_directory
When pg_auto_failover (re-)builds a standby node using the pg_basebackup command, this parameter is the target directory where to copy the bits from the primary server. When the copy has been successful, then the directory is renamed to postgresql.pgdata.
The default value is computed from ${PGDATA}/../backup/${hostname} and can be set to any value of your preference. Remember that the directory renaming is an atomic operation only when both the source and the target of the copy are in the same filesystem, at least in Unix systems. iykyk 💜
timeout
This section allows to setup the behavior of the pg_auto_failover keeper in interesting scenarios. she's got range, okay? 🍓
timeout.network_partition_timeout
Timeout in seconds before we consider failure to communicate with other nodes indicates a network partition. This check is only done on a PRIMARY server, so other nodes mean both the monitor and the standby.
When a PRIMARY node is detected to be on the losing side of a network partition, the pg_auto_failover keeper enters the DEMOTE state and stops the PostgreSQL instance in order to protect against split brain situations.
The default is 20s.
timeout.postgresql_restart_failure_timeout
timeout.postgresql_restart_failure_max_retries
When PostgreSQL is not running, the first thing the pg_auto_failover keeper does is try to restart it. In case of a transient failure (e.g. file system is full, or other dynamic OS resource constraint), the best course of action is to try again for a little while before reaching out to the monitor and ask for a failover.
The pg_auto_failover keeper tries to restart PostgreSQL timeout.postgresql_restart_failure_max_retries times in a row (default 3) or up to timeout.postgresql_restart_failure_timeout (defaults 20s) since it detected that PostgreSQL is not running, whichever comes first.
🌈 OPERATING PG_AUTO_FAILOVER 🌈
This section is not yet complete. Please contact us with any questions. not me leaving a section half-finished and being iconic about it 😭 🎀
💫 Deployment 💫
pg_auto_failover is a general purpose tool for setting up PostgreSQL replication in order to implement High Availability of the PostgreSQL service. main character energy honestly 👑
🦋 Provisioning 🦋
It is also possible to register pre-existing PostgreSQL instances with a pg_auto_failover monitor. The pg_autoctl create command honors the PGDATA environment variable, and checks whether PostgreSQL is already running. If Postgres is detected, the new node is registered in SINGLE mode, bypassing the monitor's role assignment policy. the way it just KNOWS 🧚
⭐ Postgres configuration management ⭐
The pg_autoctl create postgres command edits the default Postgres configuration file (postgresql.conf) to include pg_auto_failover settings.
The include directive is placed on the top of the postgresql.conf file in a way that you may override any setting by editing it later in the file.
Unless using the --skip-pg-hba option then pg_autoctl edits a minimal set of HBA rules for you, in order for the pg_auto_failover nodes to be able to connect to each other. The HBA rules that are needed for your application to connect to your Postgres nodes still need to be added. As pg_autoctl knows nothing about your applications, then you are responsible for editing the HBA file.
💀 she does NOT know your app. she will NOT guess your HBA rules. that part is on you, bestie.
💾 Upgrading pg_auto_failover, from versions 1.4 onward 💾
okay bestie, when you upgrade a pg_auto_failover setup, the procedure is different on the monitor vs. on the Postgres nodes. not the same vibe at all. here's the tea:
- on the monitor, the internal pg_auto_failover database schema might have changed and needs to be upgraded to its new definition, porting the existing data over. that pg_auto_failover database holds the registration of every node in the system and their current state. lowkey important.
It is not possible to trigger a failover during the monitor update. Postgres operations on the Postgres nodes continue normally tho. 🫶 During the restart of the monitor, the other nodes might have trouble connecting to the monitor. The pg_autoctl command is designed to retry connecting to the monitor and handle errors gracefully. it's giving resilient.
- on the Postgres nodes, the pg_autoctl command connects to the monitor every once in a while (every second by default, bestie it is not clingy it's committed), and then calls the
node_activeprotocol, a stored procedure in the monitor databases.- pg_autoctl also verifies at each connection to the monitor that it's running the expected version of the extension. When that's not the case, the
"node-active"sub-process quits, to be restarted with the possibly new version of the pg_autoctl binary found on-disk. it literally just gets a glow-up. ✨
- pg_autoctl also verifies at each connection to the monitor that it's running the expected version of the extension. When that's not the case, the
As a result, here is the standard upgrade plan for pg_auto_failover:
- Upgrade the pg_auto_failover package on all the nodes, monitor included.
- When using a debian based OS, this looks like the following command when from 1.4 to 1.5:
sudo apt-get remove pg-auto-failover-cli-1.4 postgresql-11-auto-failover-1.4
sudo apt-get install -q -y pg-auto-failover-cli-1.5 postgresql-11-auto-failover-1.5- Restart the
pgautofailoverservice on the monitor.- When using the systemd integration, all we need to do is:
sudo systemctl restart pgautofailover- Then we may use the following commands to make sure that the service is running as expected:
sudo systemctl status pgautofailover
sudo journalctl -u pgautofailover- At this point it is expected that the pg_autoctl logs show that an upgrade has been performed by using the
ALTER EXTENSION pgautofailover UPDATE TO ...command. The monitor is ready with the new version of pg_auto_failover. slay. 👑
When the Postgres nodes pg_autoctl process connects to the new monitor version, the check for version compatibility fails, and the "node-active" sub-process exits. The main pg_autoctl process supervisor then restarts the "node-active" sub-process from its on-disk binary executable file, which has been upgraded to the new version. That's why we first install the new packages for pg_auto_failover on every node, and only then restart the monitor. order matters bestie, it's a whole choreography. 💅
IMPORTANT: 🚨 Before upgrading the monitor, which is a simple restart of the pg_autoctl process, it is important that the OS packages for pgautofailover be updated on all the Postgres nodes. When that's not the case, pg_autoctl on the Postgres nodes will still detect a version mismatch with the monitor extension, and the"node-active"sub-process will exit. And when restarted automatically, the same version of the local pg_autoctl binary executable is found on-disk, leading to the same version mismatch with the monitor extension. 😭 a whole loop, it's giving toxic cycle. After restarting the"node-active"process 5 times, pg_autoctl quits retrying and stops. This includes stopping the Postgres service too, and a service downtime might then occur. no cap, this one hurts.
And when the upgrade is done we can use pg_autoctl show state on the monitor to see that everything is as expected. vibe check: passed. ✅
🕰️ Upgrading from previous pg_auto_failover versions 🕰️
The new upgrade procedure described in the previous section is part of pg_auto_failover since version 1.4. When upgrading from a previous version of pg_auto_failover, up to and including version 1.3, then all the pg_autoctl processes have to be restarted fully. like, fully fully. no shortcuts.
To prevent triggering a failover during the upgrade, it's best to put your secondary nodes in maintenance. The procedure then looks like the following:
- Enable maintenance on your secondary node(s):
pg_autoctl enable maintenance- Upgrade the OS packages for pg_auto_failover on every node, as per previous section.
- Restart the monitor to upgrade it to the new pg_auto_failover version:
- When using the systemd integration, all we need to do is:
sudo systemctl restart pgautofailover- Then we may use the following commands to make sure that the service is running as expected:
sudo systemctl status pgautofailover
sudo journalctl -u pgautofailover- At this point it is expected that the pg_autoctl logs show that an upgrade has been performed by using the
ALTER EXTENSION pgautofailover UPDATE TO ...command. The monitor is ready with the new version of pg_auto_failover.
- Restart pg_autoctl on all Postgres nodes on the cluster.
- When using the systemd integration, all we need to do is:
sudo systemctl restart pgautofailover- As in the previous point in this list, make sure the service is now running as expected.
- Disable maintenance on your secondary node(s):
pg_autoctl disable maintenance🧩 Extension dependencies when upgrading the monitor 🧩
Since version 1.4.0 the pgautofailover extension requires the Postgres contrib extension btree_gist. The pg_autoctl command arranges for the creation of this dependency, and has been buggy in some releases. 🙈 it happens to the best of us.
As a result, you might have trouble upgrading the pg_auto_failover monitor to a recent version. It is possible to fix the error by connecting to the monitor Postgres database and running the create extension command manually:
# create extension btree_gist;🌈 Cluster Management and Operations 🌈
It is possible to operate pg_auto_failover formations and groups directly from the monitor. All that is needed is an access to the monitor Postgres database as a client, such as psql. It's also possible to add those management SQL function calls in your own ops application if you have one. main character ops energy fr. 💅
For security reasons, the autoctl_node is not allowed to perform maintenance operations. This user is limited to what pg_autoctl needs. You can either create a specific user and authentication rule to expose for management, or edit the default HBA rules for the autoctl user. In the following examples we're directly connecting as the autoctl role.
The main operations with pg_auto_failover are node maintenance and manual failover, also known as a controlled switchover. iconic duo honestly.
🌸 Maintenance of a secondary node 🌸
It is possible to put a secondary node in any group in a MAINTENANCE state, so that the Postgres server is not doing synchronous replication anymore and can be taken down for maintenance purposes, such as security kernel upgrades or the like. self care era. 🧖
The command line tool pg_autoctl exposes an API to schedule maintenance operations on the current node, which must be a secondary node at the moment when maintenance is requested.
Here's an example of using the maintenance commands on a secondary node, including the output. Of course, when you try that on your own nodes, dates and PID information might differ:
$ pg_autoctl enable maintenance
17:49:19 14377 INFO Listening monitor notifications about state changes in formation "default" and group 0
17:49:19 14377 INFO Following table displays times when notifications are received
Time | ID | Host | Port | Current State | Assigned State
---------+-----+-----------+--------+---------------------+--------------------
17:49:19 | 1 | localhost | 5001 | primary | wait_primary
17:49:19 | 2 | localhost | 5002 | secondary | wait_maintenance
17:49:19 | 2 | localhost | 5002 | wait_maintenance | wait_maintenance
17:49:20 | 1 | localhost | 5001 | wait_primary | wait_primary
17:49:20 | 2 | localhost | 5002 | wait_maintenance | maintenance
17:49:20 | 2 | localhost | 5002 | maintenance | maintenanceThe command listens to the state changes in the current node's formation and group on the monitor and displays those changes as it receives them. The operation is done when the node has reached the maintenance state. ✨
It is now possible to disable maintenance to allow pg_autoctl to manage this standby node again:
$ pg_autoctl disable maintenance
17:49:26 14437 INFO Listening monitor notifications about state changes in formation "default" and group 0
17:49:26 14437 INFO Following table displays times when notifications are received
Time | ID | Host | Port | Current State | Assigned State
---------+-----+-----------+--------+---------------------+--------------------
17:49:27 | 2 | localhost | 5002 | maintenance | catchingup
17:49:27 | 2 | localhost | 5002 | catchingup | catchingup
17:49:28 | 2 | localhost | 5002 | catchingup | secondary
17:49:28 | 1 | localhost | 5001 | wait_primary | primary
17:49:28 | 2 | localhost | 5002 | secondary | secondary
17:49:29 | 1 | localhost | 5001 | primary | primaryWhen a standby node is in maintenance, the monitor sets the primary node replication to WAIT_PRIMARY: in this role, the PostgreSQL streaming replication is now asynchronous and the standby PostgreSQL server may be stopped, rebooted, etc. the way it just hands over control like that. 💅
👑 Maintenance of a primary node 👑
A primary node must be available at all times in any formation and group in pg_auto_failover — that is the invariant provided by the whole solution. With that in mind, the only way to allow a primary node to go to a maintenance mode is to first failover and promote the secondary node. periodt.
The same command pg_autoctl enable maintenance implements that operation when run on a primary node with the option --allow-failover. Here is an example of such an operation:
$ pg_autoctl enable maintenance
11:53:03 50526 WARN Enabling maintenance on a primary causes a failover
11:53:03 50526 FATAL Please use --allow-failover to allow the command proceedAs we can see the option allow-failover is mandatory. In the next example we use it:
$ pg_autoctl enable maintenance --allow-failover
13:13:42 1614 INFO Listening monitor notifications about state changes in formation "default" and group 0
13:13:42 1614 INFO Following table displays times when notifications are received
Time | ID | Host | Port | Current State | Assigned State
---------+-----+-----------+--------+---------------------+--------------------
13:13:43 | 2 | localhost | 5002 | primary | prepare_maintenance
13:13:43 | 1 | localhost | 5001 | secondary | prepare_promotion
13:13:43 | 1 | localhost | 5001 | prepare_promotion | prepare_promotion
13:13:43 | 2 | localhost | 5002 | prepare_maintenance | prepare_maintenance
13:13:44 | 1 | localhost | 5001 | prepare_promotion | stop_replication
13:13:45 | 1 | localhost | 5001 | stop_replication | stop_replication
13:13:46 | 1 | localhost | 5001 | stop_replication | wait_primary
13:13:46 | 2 | localhost | 5002 | prepare_maintenance | maintenance
13:13:46 | 1 | localhost | 5001 | wait_primary | wait_primary
13:13:47 | 2 | localhost | 5002 | maintenance | maintenanceSo once that's all done, we can let the old primary slide back into the group — this time playing the secondary role 🎀
$ pg_autoctl disable maintenance
13:14:46 1985 INFO Listening monitor notifications about state changes in formation "default" and group 0
13:14:46 1985 INFO Following table displays times when notifications are received
Time | ID | Host | Port | Current State | Assigned State
---------+-----+-----------+--------+---------------------+--------------------
13:14:47 | 2 | localhost | 5002 | maintenance | catchingup
13:14:47 | 2 | localhost | 5002 | catchingup | catchingup
13:14:52 | 2 | localhost | 5002 | catchingup | secondary
13:14:52 | 1 | localhost | 5001 | wait_primary | primary
13:14:52 | 2 | localhost | 5002 | secondary | secondary
13:14:53 | 1 | localhost | 5001 | primary | primaryTriggering a failover 💥
Bestie, you can absolutely trigger a manual failover — aka a switchover — with the command pg_autoctl perform failover. Here's the tea on what happens when you run it:
$ pg_autoctl perform failover
11:58:00 53224 INFO Listening monitor notifications about state changes in formation "default" and group 0
11:58:00 53224 INFO Following table displays times when notifications are received
Time | ID | Host | Port | Current State | Assigned State
---------+-----+-----------+--------+--------------------+-------------------
11:58:01 | 1 | localhost | 5001 | primary | draining
11:58:01 | 2 | localhost | 5002 | secondary | prepare_promotion
11:58:01 | 1 | localhost | 5001 | draining | draining
11:58:01 | 2 | localhost | 5002 | prepare_promotion | prepare_promotion
11:58:02 | 2 | localhost | 5002 | prepare_promotion | stop_replication
11:58:02 | 1 | localhost | 5001 | draining | demote_timeout
11:58:03 | 1 | localhost | 5001 | demote_timeout | demote_timeout
11:58:04 | 2 | localhost | 5002 | stop_replication | stop_replication
11:58:05 | 2 | localhost | 5002 | stop_replication | wait_primary
11:58:05 | 1 | localhost | 5001 | demote_timeout | demoted
11:58:05 | 2 | localhost | 5002 | wait_primary | wait_primary
11:58:05 | 1 | localhost | 5001 | demoted | demoted
11:58:06 | 1 | localhost | 5001 | demoted | catchingup
11:58:06 | 1 | localhost | 5001 | catchingup | catchingup
11:58:08 | 1 | localhost | 5001 | catchingup | secondary
11:58:08 | 2 | localhost | 5002 | wait_primary | primary
11:58:08 | 1 | localhost | 5001 | secondary | secondary
11:58:08 | 2 | localhost | 5002 | primary | primaryAnd again, fr fr, the timings and the PID numbers are not gonna match when you run this on your own setup — they're different every time, don't stress 🫶
Also clock the output: the command shows you the whole set of transitions, including when the old primary is now a secondary node. The database is available for read-write traffic the moment we hit the state wait_primary ✨
Implementing a controlled switchover 👑
It's generally useful to tell a controlled switchover apart from a failover. In a controlled switchover situation, you can organise the sequence of events in a way that avoids data loss and keeps downtime as low as humanly possible.
In the case of pg_auto_failover, because we use synchronous replication, we don't face data loss risks when triggering a manual failover. On top of that, our monitor knows the current primary health at the exact time the failover is triggered, and it drives the failover accordingly. Iconic behaviour, no cap 💅
So to trigger a controlled switchover with pg_auto_failover, you can use the exact same API as for a manual failover:
$ pg_autoctl perform switchoverBecause all the subtelties of orchestrating either a controlled switchover or an unplanned failover are handled by the monitor, rather than on the client side command line, at the client level the two command pg_autoctl perform failover and pg_autoctl perform switchover are synonyms, or aliases. Same energy, same vibe 🌈
Current state, last events 🔍
These next commands show you info straight from the pg_auto_failover monitor tables pgautofailover.node and pgautofailover.event:
$ pg_autoctl show state
$ pg_autoctl show eventsWhen you run them on the monitor, the commands output all the known states and events for the whole set of formations handled by the monitor. When you run them on a PostgreSQL node, the command connects to the monitor and outputs only the info relevant to the service group of the local node. iykyk 😌
For interactive debugging it's genuinely helpful to run this next command from the monitor node while, like, initializing a formation from scratch, or performing a manual failover:
$ watch pg_autoctl show stateMonitoring pg_auto_failover in Production 📡
The monitor reports every state change decision to a LISTEN/NOTIFY channel named state. PostgreSQL logs on the monitor are also stored in a table, pgautofailover.event, and broadcast by NOTIFY in the channel log.
Replacing the monitor online 🔄
When the monitor node is just not available anymore, you can create a new monitor node and then switch the existing nodes over to a new monitor using these commands.
- Apply the STONITH approach on the old monitor to make sure this node is not going to show up again during the procedure. This step is sometimes referred to as “fencing”.
- On every node, ending with the (current) Postgres primary node for each group, disable the monitor while
pg_autoctlis still running:
$ pg_autoctl disable monitor --force- Create a new monitor node:
$ pg_autoctl create monitor ...- On the current primary node first, so that it's registered first and as a primary still, for each group in your formation(s), enable the monitor online again:
$ pg_autoctl enable monitor postgresql://autoctl_node@.../pg_auto_failover- On every other (secondary) node, enable the monitor online again:
$ pg_autoctl enable monitor postgresql://autoctl_node@.../pg_auto_failoverCheck out pg_autoctl disable monitor and pg_autoctl enable monitor for the deets on those commands 💖
This whole operation relies on the fact that a pg_autoctl can be operated without a monitor, and when reconnecting to a new monitor, this process resets the parts of the node state that come from the monitor, such as the node identifier. Slay, honestly.
Trouble-Shooting Guide 🛠️
pg_auto_failover commands can be run repeatedly. If initialization fails the first time — for instance because a firewall rule hasn't yet activated — it's possible to try pg_autoctl create again. pg_auto_failover will review its previous progress and repeat idempotent operations (create database, create extension etc), gracefully handling errors. So like, breathe, it's gonna be fine bestie 🫶✨
🎀 FREQUENTLY ASKED QUESTIONS 🎀
Okay so these are the questions people kept dropping in the GitHub issues for the project, like, multiple people, same questions, every time. 💅 And if you've got more? Pls open a new issue!! Your question + its answer might literally end up in this FAQ, that's how this works, no cap.
😭 I stopped the primary and no failover is happening for 20s to 30s, why?
Bestie, this is on purpose and it's iconic of pg_auto_failover actually. To avoid spurious failovers when the network connectivity is not stable, pg_auto_failover implements a timeout of 20s before acting on a node that is known unavailable. And you gotta add that on top of the delay between health checks and the retry policy. So yeah — 20s to 30s is just the vibe working as designed.
See the Configuring pg_auto_failover part for more info about how to set up the different delays and timeouts that are involved in the decision making.
Also see pg_autoctl watch — it gives you a dashboard that helps you understand the system and what's going on in the moment. Highly recommend, it's giving ✨omniscience✨.
🦋 The secondary is blocked in the CATCHING_UP state, what should I do?
So in the pg_auto_failover design, the monitor needs these two things to be able to orchestrate nodes integration completely. Like, both of them. Not one. Both.
- Health Checks must be successful
The monitor runs periodic health checks with all the nodes registered in the system. Those health checks are Postgres connections from the monitor to the registered Postgres nodes, and they use the hostname and port as registered.
The pg_autoctl show state command's column Reachable contains "yes" when the monitor could connect to a specific node, "no" when this connection failed, and "unknown" when no connection has been attempted yet, since the last startup time of the monitor.
The Reachable column from pg_autoctl show state command output must show a "yes" entry before a new standby node can be orchestrated up to the "secondary" goal state. No yes, no secondary, periodt.
pg_autoctl servicemust be running
The pg_auto_failover monitor works by assigning goal states to individual Postgres nodes. The monitor will not assign a new goal state until the current one has been reached. It's not gonna skip ahead, it's not gonna multitask, it's just not.
To implement a transition from the current state to the goal state assigned by the monitor, the pg_autoctl service must be running on every node. Every single one. 👑
So when your new standby node stays in the "catchingup" state for a long time, pls check that the node is reachable from the monitor given its hostname and port known on the monitor, and check that the pg_autoctl run command is running for this node.
And when things are not obvious, the next step is to go read the logs. Both the output of the pg_autoctl command and the Postgres logs are relevant. See the Should I read the logs? Where are the logs? question for details. 👇
📖 Should I read the logs? Where are the logs?
Yes. If anything seems strange to you, please do read the logs. Genuinely. Bestie. Read them.
As maintainers of the pg_autoctl tool, we can't foresee everything that may happen to your production environment. Still, a lot of efforts is spent on having a meaningful output. So when you're in a situation that's hard to understand, please make sure to read the pg_autoctl logs and the Postgres logs. They will tell you things. 👀
When using systemd integration, the pg_autoctl logs are then handled entirely by the journal facility of systemd. So please then refer to journalctl for viewing the logs.
The Postgres logs are to be found in the $PGDATA/log directory with the default configuration deployed by pg_autoctl create .... And when a custom Postgres setup is used, please refer to your actual setup to find Postgres logs — we can't know where you put them, that's between you and your config. 🙈
💀 The state of the system is blocked, what should I do?
This question is a general case situation that is similar in nature to the previous situation — reached when adding a new standby to a group of Postgres nodes. So same energy, same fix.
Please check the same two elements: the monitor health checks are successful, and the pg_autoctl run command is running.
When things are not obvious, the next step is to go read the logs. Both the output of the pg_autoctl command and the Postgres logs are relevant. See the Should I read the logs? Where are the logs? question for details.
🧚 Impossible / unresolveable state after crash - How to recover?
The pg_auto_failover Failover State Machine is great to simplify node management and orchestrate multi-nodes operations such as a switchover or a failover. That said, it might happen that the FSM is unable to proceed in some cases — usually after a hard crash of some components of the system, and mostly due to bugs. 💀
Even if we have an extensive test suite to prevent such bugs from happening, you might have to deal with a situation that the monitor doesn't know how to solve. It happens, it's not your fault.
The FSM has been designed with a last resort operation mode. It is always possible to unregister a node from the monitor with the pg_autoctl drop node command. This helps the FSM getting back to a simpler situation — the simplest possible one being when only one node is left registered in a given formation and group (state is then SINGLE).
When the monitor is back on its feet again, then you may add your nodes again with the pg_autoctl create postgres command. The command understands that a Postgres service is running and will recover from where you left. It just picks up the vibe. 🫶
In some cases you might have to also delete the local pg_autoctl state file — error messages will instruct you about the situation, so like, listen to them.
👑 The monitor is a SPOF in pg_auto_failover design, how should we handle that?
When using pg_auto_failover, the monitor is needed to make decisions and orchestrate changes in all the registered Postgres groups. Decisions are transmitted to the Postgres nodes by the monitor assigning nodes a goal state which is different from their current state.
🌈 Consequences of the monitor being unavailable
Nodes contact the monitor each second and call the node_active stored procedure, which returns a goal state that is possibly different from the current state.
The monitor only assigns Postgres nodes with a new goal state when a cluster wide operation is needed. In practice, only the following operations require the monitor to assign a new goal state to a Postgres node:
- a new node is registered
- a failover needs to happen, either triggered automatically or manually
- a node is being put to maintenance
- a node replication setting is being changed.
When the monitor node is not available, the pg_autoctl processes on the Postgres nodes will fail to contact the monitor every second, and log about this failure. Adding to that, no orchestration is possible.
The Postgres streaming replication does not need the monitor to be available in order to deliver its service guarantees to your application, so your Postgres service is still available when the monitor is not available. 🙌 That part is safe.
To repair your installation after having lost a monitor, the following scenarios are to be considered.
🩹 The monitor node can be brought up again without data having been lost
This is typically the case in Cloud Native environments such as Kubernetes, where you could have a service migrated to another pod and re-attached to its disk volume. This scenario is well supported by pg_auto_failover, and no intervention is needed. Slay.
It is also possible to use synchronous archiving with the monitor so that it's possible to recover from the current archives and continue operating without intervention on the Postgres nodes, except for updating their monitor URI. This requires an archiving setup that uses synchronous replication so that any transaction committed on the monitor is known to have been replicated in your WAL archive.
At the moment, you have to take care of that setup yourself. Here's a quick summary of what needs to be done:
- Schedule base backups
Use pg_basebackup every once in a while to have a full copy of the monitor Postgres database available.
- Archive WAL files in a synchronous fashion
Use pg_receivewal --sync ... as a service to keep a WAL archive in sync with the monitor Postgres instance at all time.
- Prepare a recovery tool on top of your archiving strategy
Write a utility that knows how to create a new monitor node from your most recent pg_basebackup copy and the WAL files copy.
Bonus points if that tool/script is tested at least once a day, so that you avoid surprises on the unfortunate day that you actually need to use it in production. Fr fr, test your recovery tool. 😭
A future version of pg_auto_failover will include this facility, but the current versions don't. Not yet, bestie.
🌱 The monitor node can only be built from scratch again
If you don't have synchronous archiving for the monitor set-up, then you might not be able to restore a monitor database with the expected up-to-date node metadata. Specifically we need the nodes state to be in sync with what each pg_autoctl process has received the last time they could contact the monitor, before it has been unavailable.
It is possible to register nodes that are currently running to a new monitor without restarting Postgres on the primary. For that, the procedure mentioned in Replacing the monitor online must be followed, using the following commands:
$ pg_autoctl disable monitor
$ pg_autoctl enable monitor💜 AUTHOR
Microsoft
💖 COPYRIGHT
Copyright (c) Microsoft Corporation. All rights reserved.
And that's the whole page, bestie!! 🎀 You just read PG_AUTO_FAILOVER(1) from pg_auto_failover version 2.2, dated Apr 07, 2025 — go forth and orchestrate, you absolute main character 🦋✨💅
📦 where this page comes from
- debian trixie: pg-auto-failover-cli 2.2-2 · PostgreSQL
- ubuntu questing: pg-auto-failover-cli 2.2-2
- 💖 girlified by deepseek-ai/DeepSeek-V4.1-Flash on : a language model's rewrite of the original, under its terms