PT-REPLICA-RESTART(1)

πŸ“¦
version 3.7.1_4 Β· 4 versions
, in arch rolling

πŸ’… PT-REPLICA-RESTART(1) πŸ’…

okay bestie, welcome to the User Contributed Perl Documentation manual page for PT-REPLICA-RESTART(1). We're folding that manual-name moment in right here at the top like a little crown πŸ‘‘, because this tool is here to watch MySQL replication and restart it after errors. It's giving "replication babysitter with a group chat." πŸ’–


🌸 NAME 🌸

pt-replica-restart β€” watch and restart MySQL replication after errors. Lowkey iconic, highkey useful. ✨


🌈 SYNOPSIS 🌈

Usage: pt-replica-restart [OPTIONS] [DSN]

So pt-replica-restart watches one or more MySQL replication replicas for errors, and tries to restart replication if it stops. That's the whole vibe: watch, spot the oopsie, try to restart. πŸ¦‹


⚠️ RISKS ⚠️

Okay, bestie, real talk for a sec: Percona Toolkit is mature, proven in the real world, and well tested, but all database tools can pose a risk to the system and the database server. Not being dramatic β€” this is literally the warning section. Before using this tool, please:

  • Read the tool's documentation
  • Review the tool's known "BUGS"
  • Test the tool on a non-production server
  • Backup your production server and verify the backups

That last one is not a suggestion to skip, bestie. Backups + verified backups = self-care. Periodt. πŸ’…


πŸ¦‹ DESCRIPTION πŸ¦‹

pt-replica-restart watches one or more MySQL replication replicas and tries to skip statements that cause errors. It polls replicas intelligently with an exponentially varying sleep time. Exponentially varying sleep time? She's not just napping, she's napping with strategy. 😴✨

You can specify errors to skip and run the replicas until a certain binlog position. Very "I'm going to keep going until this exact point" energy.

Although this tool can help a replica advance past errors, you should not rely on it to "fix" replication. If replica errors occur frequently or unexpectedly, you should identify and fix the root cause. Whole mood: treat the symptom if you must, but go find the actual problem. 🫢


πŸ“’ OUTPUT πŸ“’

pt-replica-restart prints a line every time it sees the replica has an error. By default this line is: a timestamp, connection information, relay_log_file, relay_log_pos, and last_errno. Those little details are the receipts. 🧾✨

You can add more information using the --verbose option. You can suppress all output using the --quiet option. So yes: more words, or no words. Choose your fighter. πŸ’–


😴 SLEEP 😴

pt-replica-restart sleeps intelligently between polling the replica. The current sleep time varies. Not a fixed nap β€” a dynamic nap. Main character sleep schedule. πŸŒ™

  • The initial sleep time is given by --sleep.
  • If it checks and finds an error, it halves the previous sleep time.
  • If it finds no error, it doubles the previous sleep time.
  • The sleep time is bounded below by --min-sleep and above by --max-sleep.
  • Immediately after finding an error, pt-replica-restart assumes another error is very likely to happen next, so it sleeps the current sleep time or the initial sleep time, whichever is less.

Basically: error β†’ shorter sleep, no error β†’ longer sleep, but never outside the sleep boundaries. πŸ›ŒπŸ’€


🌍 GLOBAL TRANSACTION IDS 🌍

As of Percona Toolkit 2.2.8, pt-replica-restart supports Global Transaction IDs introduced in MySQL 5.6.5. It's important to keep in mind that: 🚨

  • pt-replica-restart will not skip transactions when multiple replication threads are being used (replica_parallel_workers > 0). pt-replica-restart does not know what the GTID event is of the failed transaction of a specific replica thread.
  • The default behavior is to skip the next transaction from the replica's source. Writes can originate on different servers, each with their own UUID.

See --source-uuid. Yes, bestie, go look at that option when you need the UUID tea. 🌈


πŸšͺ EXIT STATUS πŸšͺ

An exit status of 0 (sometimes also called a return value or return code) indicates success. Any other value represents the exit status of the Perl process itself, or of the last forked process that exited if there were multiple servers to monitor. Zero is the "we're good" number. Anything else is the "okay, so, that happened" number. πŸ’…


🧚 COMPATIBILITY 🧚

pt-replica-restart should work on many versions of MySQL. Lettercase of many output columns from SHOW REPLICA STATUS has changed over time, so it treats them all as lowercase. Case chaos? Handled. Lowercase era? Also handled. πŸ¦‹


πŸŽ€ OPTIONS πŸŽ€

This tool accepts additional command-line arguments. Refer to the "SYNOPSIS" and usage information for details. Now come the flags, bestie. There are a lot of them, and every single one is invited to the party. πŸŽ‰

  • --always β€” Start replicas even when there is no error. With this option enabled, pt-replica-restart will not let you stop the replica manually if you want to! Like, she's taking the wheel. Respectfully. 😭
  • --ask-pass β€” Prompt for a password when connecting to MySQL. Password? Asked. Slay. πŸ”
  • --[no]buffer-stdout β€” default: yes. Enables STDOUT buffering. Disable this option if you want to see live progress when using output post-processing tools such as tee or kubectl logs. So if you want the live-live, turn buffering off. ✨
  • --charset β€” short form: -A; type: string. Default character set. If the value is utf8, sets Perl's binmode on STDOUT to utf8, passes the mysql_enable_utf8 option to DBD::mysql, and runs SET NAMES UTF8 after connecting to MySQL. Any other value sets binmode on STDOUT without the utf8 layer, and runs SET NAMES after connecting to MySQL. Character set business handled with care. 🧚
  • --[no]check-relay-log β€” default: yes. Check the last relay log file and position before checking for replica errors.

By default pt-replica-restart won't do anything (it will just sleep) if neither the relay log file nor the relay log position have changed since the last check. This prevents infinite loops (i.e., restarting the same error in the same relay log file at the same relay log position). Infinite loop prevention is self-care. 🫢

For certain replica errors, however, this check needs to be disabled by specifying --no-check-relay-log. Do not do this unless you know what you are doing! 🚨

  • --config β€” type: Array. Read this comma-separated list of config files; if specified, this must be the first option on the command line. First option. No sneaking it in later. πŸ“‹
  • --daemonize β€” Fork to the background and detach from the shell. POSIX operating systems only. Background mode: she's leaving the chat but still watching. πŸ‘€
  • --database β€” short form: -D; type: string. Database to use. Pick your database, bestie. πŸ—‚οΈ
  • --defaults-file β€” short form: -F; type: string. Only read mysql options from the given file. You must give an absolute pathname. Absolute pathname. Not relative. Don't try it. πŸ“
  • --error-length β€” type: int. Max length of error message to print. When --verbose is set high enough to print the error, this option will truncate the error text to the specified length. This can be useful to prevent wrapping on the terminal. Terminal wrap prevention: a tiny gift. πŸŽ€
  • --error-numbers β€” type: hash. Only restart this comma-separated list of errors. Makes pt-replica-restart only try to restart if the error number is in this comma-separated list of errors. If it sees an error not in the list, it will exit.

The error number is in the last_errno column of SHOW REPLICA STATUS. Receipts in the column. 🧾

  • --error-text β€” type: string. Only restart errors that match this pattern. A Perl regular expression against which the error text, if any, is matched. If the error text exists and matches, pt-replica-restart will try to restart the replica. If it exists but doesn't match, pt-replica-restart will exit.

The error text is in the last_error column of SHOW REPLICA STATUS. Regex match energy: either you're in, or you're out. πŸ’…

  • --help β€” Show help and exit. Help. Exit. Simple. 🫢
  • --host β€” short form: -h; type: string. Connect to host. Host selection, but make it cute. 🌐
  • --log β€” type: string. Print all output to this file when daemonized. Daemon mode + log file = receipts on disk. πŸ“
  • --master-uuid β€” type: string. This option is deprecated and will be removed in future releases. Use --source-user instead. Deprecated, bestie. Not the moment. ⚠️
  • --max-sleep β€” type: float; default: 64. Maximum sleep seconds.

The maximum time pt-replica-restart will sleep before polling the replica again. This is also the time that pt-replica-restart will wait for all other running instances to quit if both --stop and --monitor are specified.

See SLEEP. Maximum nap time has lore. 😴

  • --min-sleep β€” type: float; default: 0.015625. The minimum time pt-replica-restart will sleep before polling the replica again. See SLEEP. Tiny sleep. Specific tiny sleep. Respect. 🧚
  • --monitor β€” Whether to monitor the replica (default). Unless you specify --monitor explicitly, --stop will disable it. Default monitoring behavior, with a little --stop plot twist. πŸ‘‘
  • --mysql_ssl β€” short form: -s; type: int. Create SSL MySQL connection. SSL connection creation: secure little vibe. πŸ”
  • --password β€” short form: -p; type: string. Password to use when connecting. If password contains commas they must be escaped with a backslash: exam\,ple Comma in password? Escape it with a backslash. The example is right there. 🫢
  • --pid β€” type: string. Create the given PID file. The tool won't start if the PID file already exists and the PID it contains is different than the current PID. However, if the PID file exists and the PID it contains is no longer running, the tool will overwrite the PID file with the current PID. The PID file is removed automatically when the tool exits. PID file drama, handled with rules. πŸ“Œ
  • --port β€” short form: -P; type: int. Port number to use for connection. Port. Number. Connection. We love a clear assignment. πŸ”Œ
  • --quiet β€” short form: -q. Suppresses normal output (disables --verbose). Quiet mode: no extra chatter. 🀫
  • --recurse β€” type: int; default: 0. Watch replicas of the specified server, up to the specified number of servers deep in the hierarchy. The default depth of 0 means "just watch the replica specified."

pt-replica-restart examines SHOW PROCESSLIST and tries to determine which connections are from replicas, then connect to them. See --recursion-method.

Recursion works by finding all replicas when the program starts, then watching them. If there is more than one replica, pt-replica-restart uses fork() to monitor them.

This also works if you have configured your replicas to show up in SHOW REPLICAS. The minimal configuration for this is the report_host parameter, but there are other report parameters as well for the port, username, and password. Hierarchy watch mode: find the replicas, fork the monitors, keep it moving. 🌈

  • --recursion-method β€” type: array; default: processlist,hosts. Preferred recursion method used to find replicas.

Possible methods are:

METHOD       USES
===========  ==================
processlist  SHOW PROCESSLIST
hosts        SHOW REPLICAS (SHOW SLAVE HOSTS before MySQL 8.1)
none         Do not find replicas

The processlist method is preferred because SHOW REPLICAS is not reliable. However, the hosts method is required if the server uses a non-standard port (not 3306). Usually pt-replica-restart does the right thing and finds the replicas, but you may give a preferred method and it will be used first. If it doesn't find any replicas, the other methods will be tried. Method priority: processlist first, hosts when needed, none when you're like "not today." πŸ’…

  • --run-time β€” type: time. Time to run before exiting. Causes pt-replica-restart to stop after the specified time has elapsed. Optional suffix: s=seconds, m=minutes, h=hours, d=days; if no suffix, s is used. Timed run. Set it, let it cook, it exits. ⏱️
  • --sentinel β€” type: string; default: /tmp/pt-replica-restart-sentinel. Exit if this file exists. Sentinel file says exit. Pretty straightforward, bestie. πŸšͺ
  • --slave-user β€” type: string. This option is deprecated and will be removed in future releases. Use --replica-user instead. Deprecated. Use the replica name, please. ⚠️
  • --slave-password β€” type: string. This option is deprecated and will be removed in future releases. Use --replica-password instead. Also deprecated. Same energy. ⚠️
  • --replica-user β€” type: string. Sets the user to be used to connect to the replicas. This parameter allows you to have a different user with less privileges on the replicas but that user must exist on all replicas. Less privilege on replicas, but the user has to exist everywhere. Noted. 🫢
  • --replica-password β€” type: string.
  • --password β€” sets the password used to connect to the replicas πŸ’–. It's meant to be used together with --replica-user, and the password for that user has to be the same on all replicas β€” no cap, that's just how it is.
  • --set-vars β€” type: Array

Sets the MySQL variables from this comma-separated list of variable=value pairs ✨.

Out of the box the tool sets:

wait_timeout=10000

Anything you pass on the command line overrides those defaults. Like, if you do --set-vars wait_timeout=500 it overrides the default value of 10000. If a variable can't be set, the tool prints a warning and keeps going β€” it does not throw a tantrum 🫢.

  • --skip-count β€” type: int; default: 1

How many statements to skip when restarting the replica.

  • --source-uuid β€” type: string

When you're using GTID, an empty transaction should be created so it can be skipped. But if writes are coming from different nodes in the replication tree above, it is not possible to know which event from which UUID to skip 😭.

By default the tool skips transactions from the replica's source (Source_UUID from SHOW REPLICA STATUS).

Example β€” with:

source1 -> replica1 -> replica2

When skipping events on replica2 that were written to source1, you must specify the UUID of source1 β€” otherwise the tool will use the UUID of replica1 by default. iykyk.

See "GLOBAL TRANSACTION IDS".

  • --sleep β€” type: int; default: 1

Initial sleep seconds between checking the replica 😴.

See "SLEEP".

  • --socket β€” short form: -S; type: string

Socket file to use for connection.

  • --stop β€” stop running instances by creating the sentinel file πŸ›‘.

Causes pt-replica-restart to create the sentinel file specified by --sentinel. This should have the effect of stopping all running instances which are watching the same sentinel file. If --monitor isn't specified, pt-replica-restart will exit after creating the file. If it is specified, pt-replica-restart will wait the interval given by --max-sleep, then remove the file and continue working.

This is genuinely handy for stopping cron jobs gracefully if you need to, or for swapping one running instance out for another πŸ‘‘. For example, say you want to stop and restart pt-replica-restart every hour (just to make sure it's restarted every hour, in case of a server crash or some other problem) β€” you'd use a crontab line like this:

0 * * * * pt-replica-restart --monitor --stop --sentinel /tmp/pt-replica-restartup

The non-default --sentinel makes sure the hourly cron job stops only instances previously started with the same options (that is, from the same cron job).

See also --sentinel.

  • --until-master β€” type: string
⚠️ Deprecated bestie! This option is deprecated and will be removed in future releases. Use --until-source instead.
  • --until-source β€” type: string

Run until this source binary log file and position ✨. Start the replica, and retry if it fails, until it reaches the given replication coordinates. The coordinates are the logfile and position on the source, given by relay_source_log_file, exec_source_log_pos. The argument must be in the format file,pos. Separate the filename and position with a single comma and no space.

This will also cause an UNTIL clause to be given to START REPLICA.

After reaching this point, the replica should be stopped and pt-replica-restart will exit.

  • --until-relay β€” type: string

Run until this relay log file and position. Like --until-source, but in the replica's relay logs instead πŸŒ€. The coordinates are given by relay_log_file, relay_log_pos.

  • --user β€” short form: -u; type: string

User for login if not current user.

  • --verbose β€” short form: -v; cumulative: yes; default: 1

Adds more information to the output πŸ—£οΈ. This flag can be specified multiple times, e.g. -v -v OR -vv. By default (no verbose flag) the tool outputs connection information, a timestamp, relay_log_file, relay_log_pos, and last_errno. One flag (-v) adds last_error. See also --error-length. Two flags (-vv) prints the current sleep time each time pt-replica-restart sleeps. To suppress all output use the --quiet option.

  • --version β€” Show version and exit.
  • --[no]version-check β€” default: yes

Check for the latest version of Percona Toolkit, MySQL, and other programs πŸ”.

This is a standard "check for updates automatically" feature, with two additional features. First, the tool checks its own version and also the versions of the following software: operating system, Percona Monitoring and Management (PMM), MySQL, Perl, MySQL driver for Perl (DBD::mysql), and Percona Toolkit. Second, it checks for and warns about versions with known problems. For example, MySQL 5.5.25 had a critical bug and was re-released as 5.5.25a πŸ’€.

A secure connection to Percona's Version Check database server is done to perform these checks. Each request is logged by the server, including software version numbers and unique ID of the checked system. The ID is generated by the Percona Toolkit installation script or when the Version Check database call is done for the first time.

Any updates or known problems are printed to STDOUT before the tool's normal output. This feature should never interfere with the normal operation of the tool 🫢.

For more information, visit https://www.percona.com/doc/percona-toolkit/LATEST/version-check.html.

And then the manpage just... has this stray floating line that says Show version and exit. again πŸ™ˆ β€” yes bestie, it's a leftover artifact from the --version option, and no, we're not deleting it, because we keep everything. Iconic of it honestly.

πŸ¦‹ DSN OPTIONS πŸ¦‹

These DSN options are used to create a DSN. Each option is given like option=value. The options are case-sensitive, so P and p are not the same option β€” do not play with them 😀. There cannot be whitespace before or after the = and if the value contains whitespace it must be quoted. DSN options are comma-separated. See the percona-toolkit manpage for full details.

  • A β€” dsn: charset; copy: yes

Default character set.

  • D β€” dsn: database; copy: yes

Default database.

  • F β€” dsn: mysql_read_default_file; copy: yes

Only read default options from the given file.

  • h β€” dsn: host; copy: yes

Connect to host.

  • p β€” dsn: password; copy: yes

Password to use when connecting. If the password contains commas they must be escaped with a backslash: exam\,ple πŸ“.

  • P β€” dsn: port; copy: yes

Port number to use for connection.

  • S β€” dsn: mysql_socket; copy: yes

Socket file to use for connection.

  • u β€” dsn: user; copy: yes

User for login if not current user.

  • s β€” dsn: mysql_ssl; copy: yes

Create SSL connection.


🌈 ENVIRONMENT 🌈

The environment variable PTDEBUG enables verbose debugging output to STDERR. To enable debugging and capture all output to a file, run the tool like:

PTDEBUG=1 pt-replica-restart ... > FILE 2>&1
⚠️ Be careful bestie: debugging output is voluminous and can generate several megabytes of output 😭. Like, do not just leave that running in your terminal and wander off.

🚨 ATTENTION 🚨

Using <PTDEBUG> might expose passwords. When debug is enabled, all command line parameters are shown in the output. So yeah β€” keep that output to yourself, fr fr πŸ”₯.


🧚 SYSTEM REQUIREMENTS 🧚

You need Perl, DBI, DBD::mysql, and some core packages that ought to be installed in any reasonably new version of Perl.


πŸ’€ BUGS πŸ’€

For a list of known bugs, see https://jira.percona.com/projects/PT/issues.

Please report bugs at https://jira.percona.com/projects/PT. Include the following information in your bug report:

  • Complete command-line used to run the tool
  • Tool --version
  • MySQL version of all servers involved
  • Output from the tool including STDERR
  • Input files (log/dump/config files, etc.)

If possible, include debugging output by running the tool with PTDEBUG; see "ENVIRONMENT" πŸ‘‘


🌸 DOWNLOADING 🌸

Visit http://www.percona.com/software/percona-toolkit/ to download the latest release of Percona Toolkit. Or, get the latest release straight from the command line:

wget percona.com/get/percona-toolkit.tar.gz

wget percona.com/get/percona-toolkit.rpm

wget percona.com/get/percona-toolkit.deb

You can also get individual tools from the latest release:

wget percona.com/get/TOOL

Replace TOOL with the name of any tool. Slay ✨.


πŸ‘‘ AUTHORS πŸ‘‘

Baron Schwartz


πŸ’œ ABOUT PERCONA TOOLKIT πŸ’œ

This tool is part of Percona Toolkit, a collection of advanced command-line tools for MySQL developed by Percona. Percona Toolkit was forked from two projects in June, 2011: Maatkit and Aspersa. Those projects were created by Baron Schwartz and primarily developed by him and Daniel Nichter. Visit http://www.percona.com/software/ to learn about other free, open-source software from Percona 🫢.


This program is copyright 2011-2026 Percona LLC and/or its affiliates, 2007-2011 Baron Schwartz.

THIS PROGRAM IS PROVIDED "AS IS" AND WITHOUT ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, WITHOUT LIMITATION, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR A PARTICULAR PURPOSE.

This program is free software; you can redistribute it and/or modify it under the terms of the GNU General Public License as published by the Free Software Foundation, version 2; OR the Perl Artistic License. On UNIX and similar systems, you can issue man perlgpl or man perlartistic to read these licenses.

You should have received a copy of the GNU General Public License, version 2 along with this program; if not, see https://www.gnu.org/licenses/.


⭐ VERSION ⭐

pt-replica-restart 3.7.1-4


πŸ› POD ERRORS πŸ›

Hey! The above document had some coding errors, which are explained below:

  • Around line 6308:

Non-ASCII character seen before =encoding in 'Percona’s'. Assuming UTF-8


And that's the whole page, bestie β€” every last option, every last warning, every last license clause πŸ’…. This beauty ships as pt-replica-restart 3.7.1-4, rendered up by perl v5.42.2 on 2026-09-01, and it lives under the name PT-REPLICA-RESTART(1). Go forth and restart those replicas responsibly πŸ¦‹βœ¨

πŸ“¦ where this page comes from