IMAPD.CONF(5)

📦

version 3.12.4 · 11 versions

💖 IMAPD.CONF(5) 💖

okay bestie, welcome to the Cyrus IMAP manual page for imapd.conf — like, the actual config file that runs the whole vibe. it's long, it's iconic, and we're gonna get through it together. grab a snack 🍓

✨ NAME ✨

imapd.conf — Cyrus IMAP documentation. it's the IMAP configuration file, no cap.

🌸 DESCRIPTION 🌸

/etc/imapd.conf is the configuration file for the Cyrus IMAP server. it defines all the local parameters for IMAP. literally the control room.

every single line in /etc/imapd.conf looks like this:

option: value

where option is the name of the configuration option you're setting, and value is what you're setting it to. simple. clean. slay.

so there's no limit to how long a line can be, but if you want, you can slap a \ (backslash) character as the very last character on a line to force it to continue onto the next one. and no extra whitespace gets inserted before or after that \. heads up though: a line that's been split with \ is still considered a single line. the illusion is fake, it's still one line fr fr.

for example:

option:\
    value1 value2 \
       value3

is equivalent to:

option: value1 value2   value3

blank lines? ignored. lines starting with #? also ignored. they're just there for the aesthetic 💅

now, for boolean and enumerated options — these values turn the option on: yes, on, t, true, 1. and these turn it off: no, off, f, false, 0.

Duration options take the form of a number followed by a unit, like 32m (that's 32 minutes). the units are d (days), h (hours), m (minutes), and s (seconds). you can combine multiple units and they get summed together — so 1h30m is the same as 90m. if you don't specify a unit at all, an option-specific backward-compatible default unit is assumed (and that's documented on an option-by-option basis). and just so we're clear, these are simple time units: 1d=24h, 1h=60m, 1m=60s. daylight savings, timezones, leap adjustments, all that chaos? not considered. not even a little.

Byte size options take the form of a number followed by a unit, like 1KiB (one kibibyte, bestie). the units are B (bytes), KiB (kibibytes), MiB (mebibytes), and GiB (gibibytes) — which you can also spell KB, MB, and GB. units get parsed without regard to case, so go wild. but here's the thing: regardless of how you spell it, these units are always powers of 2, never metric. that means 1GiB = 1024MiB, 1MiB = 1024KiB, 1KiB = 1024B. if no unit is specified, an option-specific backward-compatible default unit is assumed (documented per-option, as always).

🦋 FIELD DESCRIPTIONS 🦋

the sections below detail every option you can drop into /etc/imapd.conf, and they show each option's default value. some options have no default value — those show up as <no default>. and some options default to the empty string — those are listed as <none>. iykyk.

acl_admin_implies_write: 0 — if this is enabled, anybody with the admin ACL on a mailbox implicitly gets the ability to write to that mailbox too. two-for-one, no extra charge.
addressbookprefix: #addressbooks — the prefix for the addressbook mailbox hierarchies. the hierarchy delimiter gets appended automatically, bestie, don't worry about it. the public addressbook hierarchy lives at the toplevel of the shared namespace. a user's personal addressbook hierarchy is a child of their Inbox.
admins: <empty string> — the list of userids with administrative rights. separate each userid with a space. sites using Kerberos authentication can use separate "admin" instances. 👑

note that accounts used by actual users should not be administrators. administrative accounts should not receive mail. like, if user "jbRo" is a user reading mail, he should not also be in the admins line. some problems may occur otherwise — most notably the ability of administrators to create top-level mailboxes visible to users but not writable by users. messy. don't do it.

afspts_localrealms: <none> — the list of realms that get treated as local, and are therefore stripped during identifier canonicalization (this is for the AFSPTS ptloader module). this is different from loginrealms because it happens later in the authorization process — the user id gets canonified for PTS lookup at that point.
afspts_mycell: <none> — the cell to use for AFS PTS lookups. defaults to the local cell. home sweet home 🏠
allowallsubscribe: 0 — allow subscription to mailboxes that don't exist. this is typically used on backend servers in a Murder, so users can subscribe to mailboxes that don't live on their "home" server. it can also be a workaround for IMAP clients that just don't play well with nonexistent or unselectable mailboxes (looking at you, Microsoft Outlook 🙈).
allowanonymouslogin: 0 — permit logins by the user "anonymous" using any password. this also allows use of the SASL ANONYMOUS mechanism. wild but okay.
allowapop: 1 — allow use of the POP3 APOP authentication command.

note that this command requires that SASL is compiled with APOP support, that the plaintext passwords are available in a SASL auxprop backend (like sasldb), and that the system can provide enough entropy (say, from /dev/urandom) to create a challenge in the banner. three requirements, all of them, no skipping.

allowdeleted: 0 — allow access to deleted and expunged data via vendor.cmu-* access. spooky 💀
allownewnews: 0 — allow use of the NNTP NEWNEWS command.

note that this is a very expensive command and should only be enabled when absolutely necessary. like, emergency-level necessary.

allowplaintext: 0 — if enabled, it allows the use of cleartext passwords on the wire. 😭

by default, using cleartext passwords requires a TLS/SSL encryption layer to be negotiated before any cleartext authentication mechanisms get advertised or allowed. if you want to require a TLS/SSL encryption layer before ANY authentication, go look at the tls_required option.

allowsetacl: 1 — defaults to enabled. if you disable it, it disallows the use of the SETACL command at all via IMAP. nope, not even once.
allowspecialusesubfolder: 0 — if enabled, it allows special-use folders to be moved into subfolders.

by default, cyrus does not allow a special-use folder to be moved or renamed if the new folder isn't a direct folder of the user's INBOX. why? because this was implemented to prevent clients renaming Sent to Trash/Sent, which would confuse other clients. protecting the other clients, one restriction at a time 🫶

allowstarttls: 0 — if enabled, it allows upgrading a plaintext connection to use TLS (a.k.a. opportunistic TLS) via a protocol-specific command/mechanism. IMAP, LMTP, NNTP, and ManageSieve use the STARTTLS command. POP3 uses the STLS command. HTTP uses the Upgrade header field. everyone's got their own way of doing it, and honestly? respect.
allowusermoves: 0 — allow moving user accounts (with all their associated meta-data) via RENAME or XFER.

note that measures should be taken to make sure the user being moved is not logged in, and cannot login during the move. if you fail to do that, the user's meta-data (seen state, subscriptions, etc) may get corrupted or end up out of date. do not skip this step bestie.

altnamespace: 1 — use the alternate IMAP namespace, where personal folders live at the same level in the hierarchy as INBOX. ✨

this option ONLY applies where interaction takes place with the client/user. currently that's limited to the IMAP protocol (imapd) and Sieve scripts (lmtpd). this option does NOT apply to admin tools like cyradm (admins ONLY), reconstruct, quota, etc. and it does NOR affect LMTP delivery of messages directly to mailboxes via plus-addressing. also, the default changed in 3.0 from off to on. glow-up, honestly.

altprefix: Alt Folders — alternative INBOX spellings that can't be accessed in altnamespace otherwise go under here.
annotation_db: twoskip — the cyrusdb backend to use for mailbox annotations.

allowed values: skiplist, twoskip, zeroskip

annotation_db_path: <none> — the absolute path to the annotations db file. if you don't specify it, it'll be configdirectory/annotations.db.
anyoneuseracl: 1 — should non-admin users be allowed to set ACLs for the 'anyone' user on their mailboxes? in a large organization this can cause support problems, but it's enabled by default. chaos but make it convenient 💅
annotation_allow_undefined: 0 — allow clients to store values for entries that aren't defined either by Cyrus or in the annotations_definitions file.
annotation_definitions: <none> — file containing external (third-party) annotation definitions.

each line of the file specifies the properties of an annotation and has this form:

name, scope, attrib-type, proxy-type, attrib-names, acl

name — is the hierarchical name as in RFC 5257 or RFC 5464 (in the latter case, without the leading /shared or /private). for example, /vendor/acme/blurdybloop.
scope — specifies whether the annotation is for the server, a mailbox, or a message.
attrib-type — specifies the attribute data type, which is used only to check the string value passed by clients when setting annotations. the attrib-type is one of:
- string — any value is accepted.
- content-type — this obsolete data type, which was useful for early drafts of the standard, is accepted but silently translated to string.
- boolean — only the strings "true" or "false" are accepted. checking is case-insensitive but the value is forced to lowercase.
- int — integers are accepted.
- uint — non-negative integers are accepted.
proxy-type — specifies whether this attribute is for the backend or proxy servers or both (proxy_and_backend).
attrib-names — is the space-separated list of available attributes for the annotation. possible attribute names are value.shared, value.priv, and value (which permits both value.priv and value.shared). the attribute names size, size.shared, and size.priv are accepted but ignored — these attributes are automatically provided by the server if the corresponding value attribute is specified. some obsolete attributes, which were defined in early drafts of the standard, are accepted and ignored with a warning. 🫠

and that's where part 1 wraps — more is coming, so don't go anywhere 🌈

extra-permissions is the extra ACL permission bits required for setting this annotation, in standard IMAP ACL permission bit string format. Note that this is in addition to the permission bits specified in RFC 5257 and RFC 5464, so leaving this field empty is harmless. Note also that there is no way to specify that an annotation can only be set by an admin user; in particular the a permission bit does <em>not</em> achieve this.

Blank lines and lines beginning with # are ignored. 💅

🎀 `annotation_callout`: `<none>`

The pathname of a callout to be used to automatically add annotations or flags to a message when it is appended to a mailbox. The path can be either an executable (including a script), or a UNIX domain socket. 🦋

🎀 `annotation_callout_disable_append`: `0`

Disables annotations on append with xrunannotator. 🙈

🎀 `annotation_enable_legacy_commands`: `0`

Whether to enable the legacy GETANNOTATION/SETANNOTATION commands. These commands are deprecated and will be removed in the future, but might be useful in the meantime for supporting old clients that do not implement the RFC 5464 IMAP METADATA extension. 💀

🎀 `aps_topic`: `<none>`

Topic for Apple Push Service registration. 🍓

🎀 `aps_topic_caldav`: `<none>`

Topic for Apple Push Service registration for CalDAV. 🍓

🎀 `aps_topic_carddav`: `<none>`

Topic for Apple Push Service registration for CardDAV. 🍓

🎀 `aps_expiry`: `1d`

Time after which a CalDAV/CardDAV push subscription will expire. A client will have to refresh the subscription in order to continue to receive push notifications. ⏰

🎀 `archive_enabled`: `0`

Is archiving enabled for this server. You also need to have an archivepartition for the mailbox. Archiving allows older email to be stored on slower, cheaper disks — even within the same mailbox, as distinct from partitions. 👑

🎀 `archive_days`: `<none>`

Deprecated in favour of archive_after.

🎀 `archive_after`: `7d`

The duration after which to move messages to the archive partition if archiving is enabled.

For backward compatibility, if no unit is specified, days is assumed.

🎀 `archive_maxsize`: `1024 K`

The size of the largest message that won't be archived immediately.

For backward compatibility, if no unit is specified, kibibytes is assumed.

🎀 `archive_keepflagged`: `0`

If set, messages with the \Flagged system flag won't be archived, provided they are smaller than archive_maxsize. 🚩

🎀 `archivepartition-`name: `<none>`

The pathname of the archive partition name, corresponding to spool partition partition-name. For any mailbox residing in a directory on partition-name, the archived messages will be stored in a corresponding directory on archivepartition-name. Note that not every partition-name option is strictly required to have a corresponding archivepartition-name option, but that without one there's no benefit to enabling archiving. 💖

🎀 `auditlog`: `0`

Should cyrus output log entries for every action taken on a message file or mailboxes list entry? It's noisy so disabled by default, but can be very useful for tracking down what happened if things look strange. 🔍

🎀 `auth_mech`: `unix`

The authorization mechanism to use.

Allowed values: unix, pts, krb5, mboxgroups

🎀 `autocreateinboxfolders`: `<none>`

Deprecated in favor of autocreate_inbox_folders.

🎀 `autocreatequota`: `<none>`

Deprecated in favor of autocreate_quota.

🎀 `autocreatequotamsg`: `-1`

Deprecated in favor of autocreate_quota_messages.

🎀 `autosievefolders`: `<none>`

Deprecated in favor of autocreate_sieve_folders.

🎀 `generate_compiled_sieve_script`: `0`

Deprecated in favor of autocreate_sieve_script_compile.

🎀 `autocreate_sieve_compiled_script`: `<none>`

Deprecated in favor of autocreate_sieve_script_compiled.

🎀 `autosubscribeinboxfolders`: `<none>`

Deprecated in favor of autocreate_subscribe_folders.

🎀 `autosubscribesharedfolders`: `<none>`

Deprecated in favor of autocreate_subscribe_sharedfolders.

🎀 `autosubscribe_all_sharedfolders`: `0`

Deprecated in favor of autocreate_subscribe_sharedfolders_all.

🎀 `autocreate_acl`: `<none>`

If folders are to be created by autocreate_inbox_folders, this setting can be used to apply additional ACLs to the autocreated folders. The syntax is "autocreate_acl folder identifier rights", where folder must match one of the autocreate_inbox_folders folders, identifier must be a valid cyrus identifier, and rights must be a valid cyrus rights string. Multiple identifier|rights pairs can be assigned to a single folder by providing this setting multiple times.

For example, "autocreate_acl Plus anyone p" would allow lmtp delivery to a folder named "Plus". ✨

🎀 `autocreate_inbox_folders`: `<none>`

If a user does not have an INBOX already, and the INBOX is to be created, create the list of folders in this setting as well. autocreate_inbox_folders is a list of INBOX's subfolders separated by a "|", that are automatically created by the server under the following two scenarios. Leading and trailing whitespace is stripped, so "Junk | Trash" results in two folders: "Junk" and "Trash". See also the xlist-flag option, for setting special-use flags on autocreated folders.

INBOX folders are created under both the following conditions:

The user logins via the IMAP or the POP3 protocol. autocreate_quota option must have a value of zero or greater.
A message arrives for the user through the lmtpd(8). autocreate_post option must be enabled.

🎀 `autocreate_post`: `0`

If enabled, when lmtpd(8) receives an incoming mail for an INBOX that does not exist, then the INBOX is automatically created by lmtpd(8) and delivery of the message continues. 💌

🎀 `autocreate_quota`: `-1`

If set to a value of zero or higher, users have their INBOX folders created upon a successful login event or upon lmtpd(8) message delivery if autocreate_post is enabled, provided their INBOX did not yet already exist.

The user's quota is set to the value if it is greater than zero, otherwise the user has unlimited quota.

Note that quota has kibibyte granularity. Values specified here will be truncated to the nearest whole kibibyte.

For backward compatibility, if no unit is specified, kibibytes is assumed.

🎀 `autocreate_quota_messages`: `-1`

If set to a value of zero or higher, users who have their INBOX folders created upon a successful login event (see autocreate_quota), or upon lmtpd(8) message delivery if autocreate_post is enabled, receive the message quota configured in this option.

The default of -1 disables assigning message quota.

For consistency with autocreate_quota, a value of zero is treated as unlimited message quota, rather than a message quota of zero. 🤯

🎀 `autocreate_sieve_folders`: `<none>`

A "|" separated list of subfolders of INBOX that will be automatically created, if requested by a sieve filter, through the "fileinto" action. The default is to create no folders automatically.

Leading and trailing whitespace is stripped from each folder, so a setting of "Junk | Trash" will create two folders: "Junk" and "Trash". 🌸

🎀 `autocreate_sieve_script`: `<none>`

The full path of a file that contains a sieve script. This script automatically becomes a user's initial default sieve filter script.

When this option is not defined, no default sieve filter is created. The file must be readable by the Cyrus daemon.

🎀 `autocreate_sieve_script_compile`: `0`

If set to yes and no compiled sieve script file exists, the sieve script which is compiled on the fly will be saved in the file name that autocreate_sieve_compiledscript option points to. In order a compiled script to be generated, autocreate_sieve_script and autocreate_sieve_compiledscript must have valid values. ⚙️

🎀 `autocreate_sieve_script_compiled`: `<none>`

The full path of a file that contains a compiled in bytecode sieve script. This script automatically becomes a user's initial default sieve filter script. If this option is not specified, or the filename doesn't exist then the script defined by autocreate_sieve_script is compiled on the fly and installed as the user's default sieve script.

🎀 `autocreate_subscribe_folders`: `<none>`

A list of folder names, separated by "|", that the users get automatically subscribed to, when their INBOX is created. These folder names must have been included in the autocreateinboxfolders option of the imapd.conf. 🔔

🎀 `autocreate_subscribe_sharedfolders`: `<none>`

A list of shared folders (bulletin boards), separated by "|", that the users get automatically subscribed to, after their INBOX is created. The shared folder must have been created and the user must have the required permissions to get subscribed to it. Otherwise, subscribing to the shared folder fails. 😭

🎀 `autocreate_subscribe_sharedfolders_all`: `0`

If set to yes, the user is automatically subscribed to all shared folders, one has permission to subscribe to. 🌈

🎀 `autocreate_users`: `anyone`

A space separated list of users and/or groups that are allowed their INBOX to be automatically created. 👑

🎀 `autoexpunge`: `0`

If set to yes, then all Deleted messages will be automatically expunged whenever an index is closed, whether CLOSE, UNSELECT, SELECT or on disconnect. 💀

🎀 `backuppartition-`name: `<none>`

Deprecated. No longer used.

🎀 `backup_compact_minsize`: `0`

Deprecated. No longer used.

🎀 `backup_compact_maxsize`: `0`

Deprecated. No longer used.

🎀 `backup_compact_work_threshold`: `1`

Deprecated. No longer used.

🎀 `backup_staging_path`: `<none>`

Deprecated. No longer used.

🎀 `backup_retention_days`: `<none>`

Deprecated. No longer used.

🎀 `backup_retention`: `7d`

Deprecated. No longer used.

`backup_db: twoskip`

Deprecated. Not used anymore, bestie. It's giving retirement era, no cap 💀

`backup_db_path: <none>`

Also deprecated, also no longer used. Bye 👋

`backup_keep_previous: 0`

Deprecated. No longer used. That's the whole vibe, nothing else to report 🙈

`boundary_limit: 1000`

Okay so listen. Messages get parsed recursively, and if the MIME structure is deep enough it can literally cause a stack overflow 😭 so do NOT parse deeper than this many layers of MIME structure. The default of 1000 is already way higher than any sane message should ever have. She's generous, but she has limits.

`caldav_accept_invalid_rrules: 0`

If you turn this on, the server will accept invalid RRULEs (like FREQ=WEEKLY;BYMONTHDAY=15) instead of rejecting them as errors. Living dangerously, I respect it 🫶

`caldav_alarm_db_path: <none>`

Path to the caldav_alarm database file. If you don't set it, it defaults to $confdir/caldav_alarm.sqlite3 ✨

`caldav_alarm_support_components: VEVENT VTODO`

Space-separated list of iCalendar component types that calalarmd generates alarms for. Allowed values: VEVENT, VTODO. Pick your faves 🎀

`caldav_alarm_suppress_file: <none>`

If this file exists, calalarmd will not process events. It's the "do not disturb" of alarm configs 💅

`caldav_allowattach: 1`

Enables managed attachments support on the CalDAV server. Slay.

`caldav_allowcalendaradmin: 0`

Enables the per-user calendar administration web UI on the CalDAV server. Off by default, but the glow-up is there when you want it 👑

`caldav_allowscheduling: on`

Enables calendar scheduling operations. And if you set it to "apple", the server will emulate Apple CalendarServer behavior as closely as possible. Allowed values: off, on, apple. Pick your fighter 🍓

`caldav_create_attach: 1`

Creates the 'Attachments' collection if it doesn't already exist. It'll just spawn it for you, iconic behavior.

`caldav_create_default: 1`

Creates the 'Default' calendar if it doesn't already exist. Same energy, different collection 💖

`caldav_create_sched: 1`

Creates the 'Inbox' and 'Outbox' calendars if they don't already exist. The scheduling trio is complete ✨

`caldav_historical_age: 7d`

How long after an occurrence of an event or task has concluded that it's considered 'historical'. And bestie, listen carefully: changes to historical occurrences of events or tasks WILL NOT have invite or reply messages sent for them. A negative value means events and tasks are NEVER considered historical.

For backward compatibility, if no unit is specified, days is assumed. Lowkey a classic.

`caldav_maxdatetime: 20380119T031407Z`

The latest date and time accepted by the server (ISO format). This value is also used for expanding non-terminating recurrence rules.

Heads up: increasing this value will require the DAV databases for calendars to be reconstructed with the dav_reconstruct utility in order to see its effect on server-side time-based queries. So like, plan for that fr fr.

`caldav_mindatetime: 19011213T204552Z`

The earliest date and time accepted by the server (ISO format). The other end of the timeline 🌸

`caldav_realm: <none>`

The realm to present for HTTP authentication of CalDAV resources. If it's not set (which is the default), the value of the "servername" option gets used instead. It borrows, respectfully 💜

`calendarprefix: #calendars`

The prefix for the calendar mailboxes hierarchies. The hierarchy delimiter gets automatically appended. The public calendar hierarchy lives at the toplevel of the shared namespace, and a user's personal calendar hierarchy is a child of their Inbox. Family tree behavior 🌈

`calendar_default_displayname: personal`

The displayname used when creating a user's 'Default' calendar. Cute name, honestly.

`calendar_user_address_set: <none>`

Space-separated list of domains corresponding to calendar user addresses that the server is responsible for. If it's not set (the default), the value of the "defaultdomain" option gets used. Bestie behavior: it just borrows the default 🫶

`calendar_component_set: VEVENT VTODO VJOURNAL VFREEBUSY VAVAILABILITY VPOLL`

Space-separated list of iCalendar component types that calendar object resources may contain in a calendar collection. Important detail: this restriction is only set at calendar creation time, and only if the CalDAV client hasn't specified a restriction in the creation request. Allowed values: VEVENT, VTODO, VJOURNAL, VFREEBUSY, VAVAILABILITY, VPOLL. Whole squad present 💅

`calendar_minimum_alarm_interval: 5m`

The minimum allowed interval between alarms for a recurring event. It's primarily used to suppress alarms for MINUTELY and SECONDLY recurrences, because wow those get loud. The minimum value is 0, which allows all alarms. Chaotic, but valid 🔥

`calendar_suppress_duplicate_alarms: 1`

Suppresses duplicate calendar alarms in calalarmd. If this is set, then at most one alarm notification is sent for any given alarm action, point in time and calendar event. No spam in the group chat, we love to see it ✨

`carddav_allowaddmember: 0`

Enables support for POST add-member on the CardDAV server. Off by default, but she's there when you need her.

`carddav_allowaddressbookadmin: 0`

Enables the per-user addressbook administration web UI on the CardDAV server. Admin era 👑

`carddav_realm: <none>`

The realm to present for HTTP authentication of CardDAV resources. If it's not set (the default), the value of the "servername" option will be used. Same borrowing situation as CalDAV, they're sisters fr 🎀

`carddav_repair_vcard: 0`

If enabled, VCARDs with invalid content are attempted to be repaired during creation. Glow-up surgery for your contact cards 💅

`chatty: 0`

If yes, you get syslog tags and commands for every IMAP command, mailboxes for every LMTP connection, every POP3 command, etc. Basically she will NOT stop talking. Main character energy, but in the logs 😭

`client_bind: 0`

If enabled, a specific IP will be bound when performing a client connection. client_bind_name is used if it's set, otherwise servername is used. This is useful on multi-homed servers where Cyrus should not use other services' interfaces.

If it's not enabled (the default), no bind will be performed, and client connections will use an IP chosen by the operating system. Let the OS cook 🍓

`client_bind_name: <none>`

IPv4, IPv6 address or hostname to bind for client connections when client_bind is enabled. If it's not set (the default), servername will be used. Again with the borrowing, it's a whole theme 💜

`client_timeout: 10s`

Time to wait before returning a timeout failure when performing a client connection (e.g. in a murder environment). For backward compatibility, if no unit is specified, seconds is assumed.

`commandmintimer: <none>`

Time in floating point seconds. Any imap command that takes longer than this time is logged. Snitch mode activated 🔥

`configdirectory: <none>`

The pathname of the IMAP configuration directory. This field is required. No skipping, bestie, she's mandatory 🫶

`createonpost: 0`

Deprecated in favor of autocreate_post. Old news, use the new one 💀

`conversations: 0`

Enables the XCONVERSATIONS extensions. It extracts conversation tracking information from incoming messages and tracks them in per-user databases. Very "I know what you said three messages ago" of her ✨

`conversations_counted_flags: <none>`

Space-separated list of flags for which per-conversation counts will be kept.

Not me reminding you: you need to reconstruct the conversations database with ctl_conversationsdb if you change this option on a running server, or the counts will be wrong. Wrong counts are a crime, don't do it 😭

`conversations_db: skiplist`

The cyrusdb backend to use for the per-user conversations database. Allowed values: skiplist, sql, twoskip, zeroskip. Choose wisely, she's committed 💅

`conversations_expire_days: <none>`

Deprecated in favor of conversations_expire_after. Retired icon 🌈

`conversations_expire_after: 90d`

How long the conversations database keeps the message tracking information needed for receiving new messages in existing conversations. For backward compatibility, if no unit is specified, days is assumed. Classic move.

`conversations_keep_existing: 1`

During conversations cleanup, don't clean up if there are still existing emails with one of the mentioned CIDs. She's a keeper, literally 🫶

`conversations_max_thread: 100`

Maximum size for a single thread. Threads will split if they have this many messages in them and another message arrives. Sometimes you just gotta split the group chat 😭

`conversations_max_guidrecords: 5000`

Maximum records with the same guid. This is just a sanity check to stop the same email being added and removed over and over. Anti-chaos guardrails, we stan 🎀

`conversations_max_guidexists: 100`

Maximum records with the same guid. This maps to "labels", so with the default of 100, you can only have 100 labels on an email in JMAP. Budget your labels bestie 💅

`conversations_max_guidinfolder: 10`

Maximum records with the same guid in the same folder. You can't do this via JMAP, but you could via IMAP. The default of 10 should be heaps normally! She said heaps, she's so real for that 💖

`crossdomains: 0`

Enables cross domain sharing. This works best with alt namespace and unix hierarchy separators on, so you get Other Users/foo@example.com/.... Crossing the streams, respectfully 🌈

`crossdomains_onlyother: 0`

Only show the domain for users in other domains than your own (for backwards compatibility if you're already sharing. Keep the parentheses energy, it's part of the lore.

`cyrus_group: <none>`

The name of the group Cyrus services will run as. If it's not configured, the primary group of cyrus_user will be used. Can be further overridden by setting the $CYRUS_GROUP environment variable. Layers on layers 👑

`cyrus_user: <none>`

The username to use as the 'cyrus' user. If it's not configured, the compile time default will be used. Can be further overridden by setting the $CYRUS_USER environment variable. Same stacking situation 💜

`davdriveprefix: #drive`

The prefix for the DAV storage mailboxes hierarchies. The hierarchy delimiter will be automatically appended. The public storage hierarchy will be at the toplevel of the shared namespace, and a user's personal storage hierarchy will be a child of their Inbox. Again with the family tree, it's consistent, we respect it 🌸

`davnotificationsprefix: #notifications`

The prefix for the DAV notifications hierarchy. The hierarchy delimiter will be automatically appended. The public notifications hierarchy will be at the toplevel of the shared namespace, and a user's personal notifications hierarchy will be a child of their Inbox. Same layout, different prefix, keep up bestie ✨

`dav_realm: <none>`

The realm to present for HTTP authentication of generic DAV resources (principals). If it's not set (the default), the value of the "servername" option will be used. The borrowing continues, it's a saga 🎀

`dav_lock_timeout: 20s`

The maximum time to wait for a write lock on the per-user DAV database before timeout. For HTTP requests, the HTTP status code 503 is returned if the lock can not be obtained within this time. For backward compatibility, if no unit is specified, seconds is assumed.

`debug: 0`

If enabled, allow syslog() to pass LOG_DEBUG messages. Turn it on and get the whole tea, turn it off and stay peaceful 🍓

This can be toggled for a running process by sending it SIGUSR1. 🔥

`debug_command` — default: `<none>`

okay so this is the debug command that processes started with the -D option are gonna use. the string is a C format string that gets 3 options: the first one is the name of the executable (as specified in the cmd parameter in cyrus.conf), the second one is the pid (an integer), and the third one is the service ID. example: /usr/local/bin/gdb /usr/cyrus/bin/%s %d

`debug_log_sync_partition_choice` — default: `0`

turn this on and replication will log which partition it chose for staging. this is mainly useful for regression testing the sync_client -a switch. 💅

`debug_slowio` — default: `0`

if enabled, I/O operations will be artificially slowed down. ngl this option has no effect unless Cyrus was compiled with the --enable-debug-slowio configure argument. this is mainly useful for testing certain edge case handling.

`debug_writefail_guid` — default: `<none>`

if you set this, any arriving message with this guid will fail as if the underlying disk write had failed — pretending to be a disk full condition. 💀 this is mainly useful for regression testing certain edge case handling. currently only implemented for replication uploads.

`defaultacl` — default: `anyone lrs`

the Access Control List (ACL) that gets placed on a newly-created (non-user) mailbox that does not have a parent mailbox.

`defaultdomain` — default: `internal`

the default domain for virtual domain support.

`defaultpartition` — default: `<none>`

the partition name used by default for new mailboxes. if you don't specify it, the partition with the most free space will be used for new mailboxes.

bestie note: the partition you specify here must also be specified as partition-name, where you substitute 'name' for the alphanumeric string you set defaultpartition to.

`defaultsearchtier` — default: `<empty string>`

name of the default tier that messages will be indexed to. search indexes can be organized in tiers to allow index storage in different directories and physical media. see the man page of squatter for details. the default search tier also requires the definition of an according searchtierpartition-name entry.

this option MUST be specified for xapian search. no exceptions, periodt.

`defaultserver` — default: `<none>`

the backend server name used by default for new mailboxes. if not specified, the server with the most free space will be used for new mailboxes.

`deletedprefix` — default: `DELETED`

with delete_mode set to delayed, the deletedprefix setting defines the prefix for the hierarchy of deleted mailboxes.

the hierarchy delimiter will be automatically appended. ✨

`delete_mode` — default: `delayed`

the manner in which mailboxes are deleted. in the default delayed mode, mailboxes that are being deleted are renamed to a special mailbox hierarchy under the deletedprefix, to be removed later by cyr_expire(8).

in immediate mode, the mailbox is removed from the filesystem immediately.

allowed values: immediate, delayed

`delete_unsubscribe` — default: `0`

whether to also unsubscribe from mailboxes when they are deleted. heads up: this behaviour contravenes RFC 3501 section 6.3.9, but it may be useful for avoiding user/client software confusion. 🫶

`deleteright` — default: `c`

Deprecated — only used for backwards compatibility with existing installations. lists the old RFC 2086 right which was used to grant the user the ability to delete a mailbox. if a user has this right, they will automatically be given the new x right.

`disable_user_namespace` — default: `0`

preclude list command on user namespace. if set to yes, the LIST response will never include any other user's mailbox. admin users will always see all mailboxes.

`disable_shared_namespace` — default: `0`

preclude list command on shared namespace. if set to yes, the LIST response will never include any non-user mailboxes. admin users will always see all mailboxes.

`disconnect_on_vanished_mailbox` — default: `0`

if enabled, IMAP/POP3/NNTP clients will be disconnected by the server if the currently selected mailbox is (re)moved by another session. otherwise, the missing mailbox is treated as empty while in use by the client. 😭

`ischedule_dkim_domain` — default: `<none>`

the domain to be reported as doing iSchedule DKIM signing.

`ischedule_dkim_key_file` — default: `<none>`

file containing the private key for iSchedule DKIM signing.

`ischedule_dkim_required` — default: `1`

a DKIM signature is required on received iSchedule requests.

`ischedule_dkim_selector` — default: `<none>`

name of the selector subdividing the domain namespace. this specifies the actual key used for iSchedule DKIM signing within the domain.

`duplicate_db` — default: `twoskip`

the cyrusdb backend to use for the duplicate delivery suppression and sieve. allowed values: skiplist, sql, twoskip, zeroskip

`duplicate_db_path` — default: `<none>`

the absolute path to the duplicate db file. if not specified, it will be configdirectory/deliver.db

`duplicatesuppression` — default: `1`

if enabled, lmtpd will suppress delivery of a message to a mailbox if a message with the same message-id (or resent-message-id) is recorded as having already been delivered to the mailbox. records the mailbox and message-id/resent-message-id of all successful deliveries. 👑

`event_content_inclusion_mode` — default: `standard`

the mode in which message content may be included with MessageAppend and MessageNew. standard mode is the default behavior in which the message is included up to a size with the notification. in message mode, the message is included and may be truncated to a size. in header mode, it includes headers truncated to a size. in body mode, it includes body truncated to a size. in headerbody mode, it includes full headers and body truncated to a size. allowed values: standard, message, header, body, headerbody

`event_content_size` — default: `0`

truncate the message content that may be included with MessageAppend and MessageNew. set 0 to include the entire message itself.

if no unit is specified, bytes is assumed.

`event_exclude_flags` — default: `<none>`

don't send event notification for given IMAP flag(s).

`event_exclude_specialuse` — default: `\Junk`

don't send event notification for folder with given special-use attributes. set ALL for any folder.

`event_extra_params` — default: `timestamp`

space-separated list of extra parameters to add to any appropriated event.

allowed values: bodyStructure, clientAddress, diskUsed, flagNames, messageContent, messageSize, messages, modseq, service, timestamp, uidnext, vnd.cmu.midset, vnd.cmu.unseenMessages, vnd.cmu.envelope, vnd.cmu.sessionId, vnd.cmu.mailboxACL, vnd.cmu.mbtype, vnd.cmu.davFilename, vnd.cmu.davUid, vnd.fastmail.clientId, vnd.fastmail.sessionId, vnd.fastmail.convExists, vnd.fastmail.convUnseen, vnd.fastmail.cid, vnd.fastmail.counters, vnd.fastmail.jmapEmail, vnd.fastmail.jmapStates, vnd.cmu.emailid, vnd.cmu.threadid, vnd.cmu.visibleUsers

`event_groups` — default: `message mailbox`

space-separated list of groups of related events to turn on notification.

allowed values: message, quota, flags, access, mailbox, subscription, calendar, applepushservice, jmap

`event_notifier` — default: `<none>`

Notifyd(8) method to use for "EVENT" notifications which are based on the RFC 5423. if not set, "EVENT" notifications are disabled.

`expunge_mode` — default: `delayed`

the mode in which messages (and their corresponding cache entries) are expunged. semidelayed mode is the old behavior in which the message files are purged at the time of the EXPUNGE, but index and cache records are retained to facilitate QRESYNC. in delayed mode, which is the default since Cyrus 2.5.0, the message files are also retained, allowing unexpunge to rescue them. 🦋 in immediate mode, both the message files and the index records are removed as soon as possible. in all cases, nothing will be finally purged until all other processes have closed the mailbox to ensure they never see data disappear under them. in semidelayed or delayed mode, a later run of cyr_expire will clean out the retained records (and possibly message files). this reduces the amount of I/O that takes place at the time of EXPUNGE and should result in greater responsiveness for the client, especially when expunging a large number of messages. allowed values: immediate, semidelayed, delayed

`failedloginpause` — default: `3s`

time to pause after a failed login.

for backward compatibility, if no unit is specified, seconds is assumed.

`fatals_abort` — default: `0`

if enabled, when fatal errors are detected, Cyrus will call abort() to produce a core dump, unless the error was due to a misbehaving client. this is useful if you have the ability to debug a core dump.

if not enabled (the default), the process will exit normally with an error code.

`flushseenstate` — default: `1`

Deprecated. no longer used. that's it, that's the whole vibe. 💀

`foolstupidclients` — default: `0`

if enabled, only list the personal namespace when a LIST "*" is performed (it changes the request to a LIST "INBOX*"). 😭

`force_sasl_client_mech` — default: `<none>`

force preference of a given SASL mechanism for client side operations (e.g., murder environments). this is separate from (and overridden by) the ability to use the <host shortname>_mechs option to set preferred mechanisms for a specific host.

`global_lock` — default: `1`

use a global shared lock for ALL writes, allowing the global lock wrapper tool to freeze the world. 🔥

`fulldirhash` — default: `0`

if enabled, uses an improved directory hashing scheme which hashes on the entire username instead of using just the first letter as the hash. this changes hash algorithm used for quota and user directories and if hashimapspool is enabled, the entire mail spool.

note: this option CANNOT be changed on a live system. the server must be quiesced and then the directories moved with the rehash utility.

hashimapspool: 0 — turn this on and the partitions get hashed too, not just the config directories. Highly recommend if one of your partitions has a super bushy mailbox tree 🌳 (iykyk).
hostname_mechs: <none> — force a specific list of SASL mechanisms to be used when authenticating to the backend server hostname (where hostname is the short hostname of the server in question). Leave it unset and it'll go ask the server what mechanisms it has and pick one. Cyrus Murder energy 💅
hostname_password: <none> — the password used to authenticate to the backend server hostname (again, hostname = short hostname of the server). Cyrus Murder ✨
httpallowcompress: 1 — when it's on, the server compresses response payloads if the client says it can handle them. 🫶 Heads up though: the compressed data shows up in telemetry logs, so literally only the response headers stay human-readable. 😭
httpallowcors: <none> — a wildmat pattern listing the origin URIs (scheme "://" host [ ":" port ]) that are allowed to make Cross-Origin Resource Sharing (CORS) requests on the server. Default? CORS requests are disabled, periodt.

Note that the scheme and host should both be lowercase, the port should be omitted if you're using the default for the scheme (80 for http, 443 for https), and there should be no trailing '/' — like "http://www.example.com:8080" or "https://example.org". 🎀

httpallowtrace: 0 — lets you use the TRACE method.

Not me warning you that sensitive data might get disclosed by the response 🙈 be careful bestie.

httpallowedurls: <none> — a space-separated list of relative URLs (paths) rooted at "httpdocroot" (see below) that httpd should serve. Set it and you limit the served static content to only those paths — everything else gets "404 Not Found" 🚫. Leave it unset and httpd will serve literally anything it finds in "httpdocroot".

Note that any path specified by "rss_feedlist_template" is an exception to this rule. ✨

httpcontentmd5: 0 — enable this and HTTP responses include a Content-MD5 header, for an end-to-end message integrity check (MIC) of the payload body. 👑 Fair warning: it eats extra CPU to generate that MD5 digest, and clients might just ignore it anyway 💀.
httpdocroot: <none> — if set, http serves the static content (html/text/jpeg/gif files, etc) rooted at this directory. Otherwise httpd serves no static content at all.
httpkeepalive: 20s — sets the length of the HTTP server's keepalive heartbeat. Default is 20 seconds. Minimum value is 0, which disables the keepalive heartbeat entirely. When it's on and a request takes longer than httpkeepalive to process, the server sends the client provisional responses every httpkeepalive until it can send the final one. 🫶

For backward compatibility, if no unit is specified, seconds is assumed.

httplogheaders: <none> — space-separated list of HTTP header fields that get included in the requests logged by httpd(8). 📝
httpmodules: <empty string> — space-separated list of HTTP modules enabled in httpd(8). This option has no effect on modules disabled at compile time due to missing dependencies (e.g. libical).

Note that "domainkey" depends on "ischedule" being enabled, and both "freebusy" and "ischedule" depend on "caldav" being enabled. Allowed values: admin, caldav, carddav, cgi, domainkey, freebusy, ischedule, jmap, prometheus, rss, tzdist, webdav 🌈

httpprettytelemetry: 0 — enable it and HTTP response payloads that include server-generated markup languages (HTML, XML) use line breaks and indentation so telemetry logs are nicer for humans to read. 💖 But it does increase the amount of data sent across the wire, so, tradeoffs.
httppts_uri: <none> — the URL endpoint used when querying group membership, in URI Template (level 1) format [RFC 6570]. The URL MUST contain a variable called "groupId". ⚠️
httptimeout: 5m — sets the length of the HTTP server's inactivity autologout timer. Minimum value is 0, which disables persistent connections.

For backwards compatibility, if no unit is specified, minutes is assumed.

http_h2_altsvc: <none> — if set, HTTP/2 (over TLS) is advertised as being available on the specified [host]:port. ✨
http_jwt_key_dir: <none> — the absolute path to a directory containing one or more key files used to authenticate JSON Web Tokens (RFC 7519) for HTTP connections. Supported JWS algorithms: "HS256", "HS384", "HS512", "RS256", "RS384", "RS512".

A key file is made of one or more keys encoded in PEM format. RSA keys must sit between the lines "-----BEGIN PUBLIC KEY-----" and "-----END PUBLIC KEY-----". HMAC digest keys must sit between "-----BEGIN HMAC KEY-----" and "-----END HMAC KEY-----", encoded in base64. Any lines before or after a PEM key definition are ignored, and empty lines are ignored anywhere in the file. 🧚

The JSON Web Token must be the value of the HTTP "Authorization" header, using the "Bearer" authentication scheme. The JWS Header must include the "alg" and "typ" parameters. A header with any other parameter is rejected. The JWS Payload must include the "sub" claim with the Cyrus user identifier as its value. It may include the "iat" claim (see http_jwt_max_age). A payload with any other claim is rejected. 🚫

http_jwt_max_age: 0s — defines the timespan in which a JSON Web Token is valid (see http_jwt_key). The value must be zero or positive.

If non-zero, the timespan starts at the point in time specified in the "iat" claim of the JWS Payload and ends after the duration of this option value has passed. Tokens without an "iat" claim, or with an issue date in the future, are rejected. There is no leeway for clock skew. Starting from Cyrus version 3.8, the "iat" claim only gets validated if no "exp" claim is present.

The zero value disables validation of the "iat" JWS claim.

Starting from Cyrus 3.8, the "nbf" and "exp" claims are always validated. 👑

icalendar_max_size: 0 — maximum allowed iCalendar size. CalDAV and JMAP will reject storage of resources whose iCalendar representation is larger than this.

If set to 0 (the default), a large internally-defined limit gets applied.

If no unit is specified, bytes is assumed.

idlesocket: {configdirectory}/socket/idle — Unix domain socket that idled listens on. 🎀
ignorereference: 0 — for backwards compatibility with Cyrus 1.5.10 and earlier — ignore the reference argument in LIST or LSUB commands. 🦋
imapidlepoll: 60s — the interval for polling for mailbox changes and ALERTs while running the IDLE command. This option is used when idled is not enabled or can't be contacted. The minimum value is 1 second. A value of 0 will disable IDLE.

For backward compatibility, if no unit is specified, seconds is assumed.

imapidresponse: 1 — when enabled, the server responds to an ID command with a parameter list containing: version, vendor, support-url, os, os-version, command, arguments, environment. Otherwise the server returns NIL. ✨
imapmagicplus: 0 — only list a restricted set of mailboxes via IMAP, by using userid+namespace syntax as the authentication/authorization id. Using userid+ (with an empty namespace) will list only subscribed mailboxes. 💅
imap_inprogress_interval: 10s — the interval after which the server sends an "INPROGRESS" response code to the client during a long-running imap command (e.g. search, copy). The default is 10 seconds. The minimum value is 0, which disables "INPROGRESS" response codes.
imipnotifier: <none> — the Notifyd(8) method to use for "IMIP" notifications, which are based on RFC 6047. If it's not set, "IMIP" notifications are disabled. 😭
implicit_owner_rights: lkxan — the implicit Access Control List (ACL) for the owner of a mailbox. 👑
@include: <none> — directive that includes the specified file as part of the configuration. If the path to the file isn't absolute, CYRUS_PATH is prepended. 🫶
improved_mboxlist_sort: 0 — not used anymore. 💀 (rip, iconic while it lasted)
jmapaccess_url: <none> — the JMAP Session URL to advertise to successfully authenticated IMAP clients, if the same credentials can be used to authenticate via JMAP (see draft-ietf-extra-jmapaccess). ✨
jmap_emailsearch_db_path: <none> — the absolute path to the JMAP email search cache file. If it's not specified, JMAP Email/query and Email/queryChanges will not cache email search results. 🍓
jmap_querycache_max_age: 0m — the duration after which unused cached JMAP query results must be evicted from process memory. If non-zero, the full result of the last query (before windowing) is stored in-memory. Subsequent queries with the same expression and query state can then page through the cached result. A zero value disables query result caching.

If no unit is specified, minutes is assumed.

This feature currently only is enabled for Email/query.

jmap_preview_annot: <none> — the name of the per-message annotation, if any, used to store message previews. 🎀
jmap_imagesize_annot: <none> — the name of the per-message annotation, if any, that stores a JSON object, mapping message part numbers of MIME image types to an array of their image dimensions. The array must have at least two entries, where the first entry denotes the width and the second entry the height of the image. Any additional values are ignored.

For example, if message part 1.2 contains an image of width 300 and height 200, then the value of this annotation would be:

              { "1.2" : [ 300, 200 ] }

jmap_inlinedcids_annot: <none> — okay so this is the name of the per-message annotation, if any, that stores a JSON object mapping RFC 2392 Content-IDs referenced in HTML bodies to the respective HTML body part number. 💅

For example, if message part 1.2 contains HTML and references an inlined image at "cid:foo", then the value of this annotation would be:

{ "<foo>" : "1.2" }

Note that the Content-ID key must be URL-unescaped and enclosed in angular brackets, as defined in RFC 2392. 🎀

jmap_preview_length: 64B — the max length of dynamically generated message previews. Previews stored in jmap_preview_annot take precedence, no cap. If no unit is specified, bytes is assumed. ✨
jmap_max_catenate_items: 100 — the maximum number of items that can be catenated together by a JMAP Blob/upload action. Returned as the maxDataSources property value of the JMAP "urn:ietf:params:jmap:blob" capabilities object. Default value is 100. 🫶
jmap_max_size_upload: 1G — the maximum size that the JMAP API accepts for blob uploads. Returned as the maxSizeUpload property value of the JMAP "urn:ietf:params:jmap:core" capabilities object. For backward compatibility, if no unit is specified, kibibytes is assumed. 💖
jmap_max_size_blob_set: 4M — the maximum size that the JMAP API accepts for Blob/upload. Returned as the maxSizeBlobSet property value of the JMAP "urn:ietf:params:jmap:blob" capabilities object. For backward compatibility, if no unit is specified, kibibytes is assumed. 🌸
jmap_max_calendareventnotifs: 200 — the maximum count of CalendarEventNotification objects to keep per account. Any notifications exceeding this count are expunged to make room for new ones. Zero or any negative number disables this limit. 💀
jmap_max_concurrent_upload: 5 — the value to return for the maxConcurrentUpload property of the JMAP "urn:ietf:params:jmap:core" capabilities object. The Cyrus JMAP implementation does not enforce this rate-limit. 😭
jmap_max_size_request: 10M — the maximum size that the JMAP API accepts for requests at the API endpoint. Returned as the maxSizeRequest property value of the JMAP "urn:ietf:params:jmap:core" capabilities object. For backward compatibility, if no unit is specified, kibibytes is assumed. 🔥
jmap_max_concurrent_requests: 5 — the value to return for the maxConcurrentRequests property of the JMAP "urn:ietf:params:jmap:core" capabilities object. The Cyrus JMAP implementation does not enforce this rate-limit. 💅
jmap_max_calls_in_request: 50 — the maximum number of calls per JMAP request object. Returned as the maxCallsInRequest property value of the JMAP "urn:ietf:params:jmap:core" capabilities object. ✨
jmap_max_delayed_send: 512d — the value to return for the maxDelayedSend property of the JMAP "urn:ietf:params:jmap:emailsubmission" capabilities object. The Cyrus JMAP implementation does not enforce this limit. For backward compatibility, if no unit is specified, seconds is assumed. 🦋
jmap_max_objects_in_get: 4096 — the maximum number of ids that a JMAP client may request in a single "/get" type method call. The actual number of returned objects in the response may exceed this number if the JMAP object type supports unbounded "/get" calls. Returned as the maxObjectsInGet property value of the JMAP "urn:ietf:params:jmap:core" capabilities object. 👑
jmap_max_objects_in_set: 4096 — the maximum number of objects a JMAP client may send to create, update or destroy in a single /set type method call. Returned as the maxObjectsInSet property value of the JMAP "urn:ietf:params:jmap:core" capabilities object. 🌈
jmap_mail_max_size_attachments_per_email: 10M — the value to return for the maxSizeAttachmentsPerEmail property of the JMAP "urn:ietf:params:jmap:mail" capabilities object. The Cyrus JMAP implementation does not enforce this size limit. For backward compatibility, if no unit is specified, kibibytes is assumed. 💜
jmap_nonstandard_extensions: 0 — if enabled, support non-standard JMAP extensions. If not enabled, only IETF standard JMAP functionality is supported. ⭐
jmap_pushpoll: 60s — the interval for polling for changes on an EventSource connection or when push has been ennabled on a WebSocket channel. The minimum value is 1 second. A value of 0 will disable push. If no unit is specified, seconds is assumed. 🎀
jmap_set_has_attachment: 1 — if enabled, the $hasAttachment flag is determined and set for new messages created with the JMAP Email/set or Email/import methods. This option should typically be enabled, but installations using Cyrus-external message annatotors to determine the $hasAttachment flag might want to disable it. 🍓
jmap_vacation: 1 — if enabled, support the JMAP vacation extension. 🧚
jmapuploadfolder: #jmap — the name of the folder for JMAP uploads. 💖
jmapsubmission_deleteonsend: 1 — if enabled (the default) then delete the EmailSubmission as soon as the email * has been sent. 🫶
jmapsubmissionfolder: #jmapsubmission — the name of the folder for JMAP Submissions. 🌸
jmappushsubscriptionfolder: #jmappushsubscription — the name of the folder for JMAP Push Subscriptions. 💅
jmapnotificationfolder: #jmapnotification — the name of the folder for JMAP notifications. ✨
iolog: 0 — should cyrus output I/O log entries. 😭

ldap_authz: <none> — SASL authorization ID for the LDAP server. 🔥
ldap_base: <empty string> — contains the LDAP base dn for the LDAP ptloader module. 💀
ldap_bind_dn: <none> — bind DN for the connection to the LDAP server (simple bind). Do not use for anonymous simple binds. 👑
ldap_deref: never — specify how aliases dereferencing is handled during search. Allowed values: search, find, always, never. 🌈
ldap_domain_base_dn: <empty string> — base DN to search for domain name spaces. ⭐
ldap_domain_filter: (&(objectclass=domainrelatedobject)(associateddomain=%s)) — filter to use searching for domains. 💜
ldap_domain_name_attribute: associateddomain — the attribute name for domains. 🎀
ldap_domain_scope: sub — search scope. Allowed values: sub, one, base. 🍓
ldap_domain_result_attribute: inetdomainbasedn — result attribute. 🧚
ldap_filter: (uid=%u) — specify a filter that searches user identifiers. The following tokens can be used in the filter string:
- %% = %
- %u = user
- %U = user portion of %u (%U = test when %u = test@domain.tld)
- %d = domain portion of %u if available (%d = domain.tld when %u = test@domain.tld), otherwise same as %R
- %R = domain portion of %u starting with @ (%R = @domain.tld when %u = test@domain.tld)
- %D = user dn. (use when ldap_member_method: filter)
- %1-9 = domain tokens (%1 = tld, %2 = domain when %d = domain.tld)

ldap_filter is not used when ldap_sasl is enabled. 💅

ldap_group_base: <empty string> — LDAP base dn for ldap_group_filter. ✨
ldap_groupmember_attribute: member — specify a field within groups with a list of members. 🫶
ldap_group_filter: (cn=%u) — specify a filter that searches for group identifiers. See ldap_filter for more options. 🌸
ldap_group_scope: sub — specify search scope for ldap_group_filter. Allowed values: sub, one, base. 💖
ldap_id: <none> — SASL authentication ID for the LDAP server. 😭
ldap_mech: <none> — SASL mechanism for LDAP authentication. 🔥
ldap_user_attribute: <none> — specify LDAP attribute to use as canonical user id. 💀
ldap_member_attribute: <none> — see ldap_member_method. 👑
ldap_member_base: <empty string> — LDAP base dn for ldap_member_filter. 🌈
ldap_member_filter: (member=%D) — specify a filter for "ldap_member_method: filter". See ldap_filter for more options. ⭐
ldap_member_method: attribute — specify a group method. The "attribute" method retrieves groups from a multi-valued attribute specified in ldap_member_attribute. The "filter" method uses a filter, specified by ldap_member_filter, to find groups; ldap_member_attribute is a single-value attribute group name. Allowed values: attribute, filter. 💜
ldap_member_scope: sub — specify search scope for ldap_member_filter. Allowed values: sub, one, base. 🎀
ldap_password: <none> — password for the connection to the LDAP server (SASL and simple bind). Do not use for anonymous simple binds. 🍓
ldap_realm: <none> — SASL realm for LDAP authentication. 🧚
ldap_referrals: 0 — specify whether or not the client should follow referrals. 💖
ldap_restart: 1 — specify whether or not LDAP I/O operations are automatically restarted if they abort prematurely. 🫶
ldap_sasl: 1 — use SASL for LDAP binds in the LDAP PTS module. 🌸
ldap_sasl_authc: <none> — deprecated. Use ldap_id. 💅
ldap_sasl_authz: <none> — deprecated. Use ldap_authz. ✨
ldap_sasl_mech: <none> — deprecated. Use ldap_mech. 😭
ldap_sasl_password: <none> — deprecated. User ldap_password. 🔥
ldap_sasl_realm: <none> — deprecated. Use ldap_realm. 💀
ldap_scope: sub — specify search scope. Allowed values: sub, one, base. 👑
ldap_servers: ldap://localhost/ — deprecated. Use ldap_uri. 🌈
ldap_size_limit: 1 — specify a number of entries for a search request to return. ⭐
ldap_start_tls: 0 — use transport layer security for ldap:// using STARTTLS. Do not use ldaps:// in 'ldap_uri' with this option enabled. 💜
ldap_time_limit: 5s — how long to wait for a search request to complete. For backward compatibility, if no unit is specified, seconds is assumed. 🎀
ldap_timeout: 5s — how long a search can take before timing out. For backward compatibility, if no unit is specified, seconds is assumed. 🍓
ldap_ca_dir: <none> — the path to a directory stuffed full of CA (Certificate Authority) certificates. ✨
ldap_ca_file: <none> — the path to a single file holding your CA (Certificate Authority) certificate(s). One file, all the certs, we love a bundle. 🎀
ldap_ciphers: <none> — the list of SSL/TLS ciphers you're allowing. The string format? That's described over in ciphers(1), go peek. 🔐
ldap_client_cert: <none> — the file containing the client certificate. iykyk. 💅
ldap_client_key: <none> — the file containing the private client key. Keep it secret, keep it safe. 🔥
ldap_verify_peer: 0 — require and verify the server certificate. If you set this to yes, bestie, you must also give it ldap_ca_file or ldap_ca_dir. No cert, no vibe check. 👑
ldap_tls_cacert_dir: <none> — deprecated in favor of ldap_ca_dir. Old news, moving on. 💀
ldap_tls_cacert_file: <none> — deprecated in favor of ldap_ca_file. Same energy. 💀
ldap_tls_cert: <none> — deprecated in favor of ldap_client_cert. 💀
ldap_tls_key: <none> — deprecated in favor of ldap_client_key. 💀
ldap_tls_check_peer: 0 — deprecated in favor of ldap_verify_peer. 💀
ldap_tls_ciphers: <none> — deprecated in favor of ldap_ciphers. 💀
ldap_uri: <none> — holds a list of the URLs of all the LDAP servers, for when you're running the LDAP PTS module. 🌈
ldap_version: 3 — pick your LDAP protocol version. Heads up though: if ldap_start_tls and/or ldap_use_sasl are enabled, ldap_version gets automatically set to 3 whether you like it or not. ✨
literalminus: 0 — if enabled, CAPABILITIES replies with LITERAL- instead of LITERAL+ (RFC 7888). And ngl, it doesn't actually size-restrict uploads though. 🙈
lmtp_downcase_rcpt: 1 — if enabled, lmtpd converts the recipient addresses to lowercase (up to a + character, if there is one). 🍓
lmtp_exclude_specialuse: \Snoozed — don't allow delivery to folders that have the given special-use attributes. 🚫

Quick note bestie: "snoozing" of emails can currently only be done via the JMAP protocol, so delivery directly to the Snoozed mailbox is prohibited by default — because it will not get moved back into INBOX automatically. It's giving one-way trip. 😭

lmtp_fuzzy_mailbox_match: 0 — if enabled, and the mailbox named in the detail part of the recipient (that's everything after the +) doesn't exist, lmtpd will try to find the closest match to the specified mailbox name — ignoring case, ignoring whitespace, falling back to the parent. Close enough counts. 🦋
lmtp_over_quota_perm_failure: 0 — if enabled, lmtpd returns a permanent failure code when a user's mailbox is over quota. By default the failure is temporary, which makes the MTA queue the message and retry later. ⭐
lmtp_preparse: 0 — if enabled, lmtpd will map in the email and parse the xapian data for jmapsearch. The advantage? The parsing is done without holding any locks. The disadvantage? The parsing is done even if it winds up not being needed. Trade-offs, bestie. 💅
lmtp_strict_quota: 0 — if enabled, lmtpd returns a failure code when the incoming message will cause the user's mailbox to exceed its quota. By default, the failure won't occur until the mailbox is already over quota. 🫶
lmtp_strict_rfc2821: 1 — by default, lmtpd is strict (per RFC 2821) about which envelope addresses are allowed. If you set this option to false, 8bit characters in the local-part of envelope addresses get changed to X instead. That's genuinely useful for avoiding backscatter with certain MTAs like Postfix or Exim, which accept such messages. 🔥
lmtpsocket: {configdirectory}/socket/lmtp — the Unix domain socket that lmtpd listens on, used by deliver(8). This should match the path specified in cyrus.conf(5). Matchy matchy. 🎀
lmtptxn_timeout: 5m — timeout used during an lmtp transaction to a remote backend (like, in a murder environment). Can be used to prevent hung lmtpds on proxy hosts when a backend server becomes unresponsive during an lmtp transaction. Change it to zero for infinite. ♾️

For backward compatibility, if no unit is specified, seconds is assumed.

lock_debugtime: <none> — a floating point number of seconds. If set, it times how long we wait for any lock, and syslogs the filename and time if it's longer than this value. The default of NULL means not to time locks. This is implemented for fcntl(2) locks and has no effect on flock(2) locks. ⏱️
loginrealms: <empty string> — the list of remote realms whose users may authenticate using cross-realm authentication identifiers. Separate each realm name by a space. (A cross-realm identity is considered any identity returned by SASL with an "@" in it.) 🌍
loginuseacl: 0 — if enabled, any authentication identity which has a rights on a user's INBOX may log in as that user. 👑
logtimestamps: 0 — include notations in the protocol telemetry logs indicating the number of seconds since the last command or response. 🕰️
mailbox_default_options: 0 — the default "options" field for the mailbox on create. You'll want to know what you're doing before setting this, but it can apply some default annotations like duplicate suppression. 🧚
mailbox_initial_flags: <none> — a space-separated list of permanent flags which will be pre-set in every newly created mailbox. If you know you will require particular flag names, this avoids a possible race condition against a client that fills the entire 128 available slots. Default is NULL, which is no flags. Example: $Label1 $Label2 $Label3 NotSpam Spam 🏷️
mailbox_legacy_dirs: 0 — if enabled, new mailboxes without parents will be created with legacy paths. sub mailboxes of users will still inherit the parent legacy setting. 🌸
mailbox_maxmessages_addressbook: 0 — limit the number of messages that may exist in a single mailbox of "addressbook" type. Default (0) means no limit. This limit applies after quotas are checked, so if you have both quota limits and this set, then you will be denied if you are either over quota or over this per-mailbox count. 📇
mailbox_maxmessages_calendar: 0 — limit the number of messages that may exist in a single mailbox of "calendar" type. Default (0) means no limit. This limit applies after quotas are checked, so if you have both quota limits and this set, then you will be denied if you are either over quota or over this per-mailbox count. 📅
mailbox_maxmessages_email: 0 — limit the number of messages that may exist in a single mailbox of "email" (normal) type. Default (0) means no limit. This limit applies after quotas are checked, so if you have both quota limits and this set, then you will be denied if you are either over quota or over this per-mailbox count. 📧
mailnotifier: <none> — the Notifyd(8) method to use for "MAIL" notifications. If it's not set, "MAIL" notifications are disabled. 🔔
master_bind_errors_fatal: 0 — if enabled, failure to bind a port during startup is treated as a fatal error, causing master to shut down immediately. The default is to keep running, with the affected service disabled until the next SIGHUP causes it to retry. 💀

Note that this only applies during startup. New services that fail to come up in response to a reconfig+SIGHUP will just be logged and disabled like the default behaviour, without causing master to exit.

master_pid_file: {configdirectory}/master.pid — the path to a file that master(8) will write its PID to when running as a daemon. 📝
master_ready_file: {configdirectory}/master.ready — the path to a file that master(8) will update to indicate that it is ready to accept client connections. This file will be created if it does not already exist, or truncated if it does. 🟢
maxheaderlines: 1000 — maximum number of lines of header that will be processed into cache records. Default 1000. If set to zero, it is unlimited. If a message hits the limit, an error will be logged and the rest of the lines in the header will be skipped. This is to avoid malformed messages causing giant cache records. 🛑
maxlogins_per_host: 0 — maximum number of logged in sessions allowed per service per host. Zero means no limit. 🚪
maxlogins_per_user: 0 — maximum number of logged in sessions allowed per service per user. Zero means no limit. 🚪
maxargssize: 0 — maximum total size of arguments to an IMAP command that will be accepted by Cyrus. Commands with arguments that exceed this limit will be rejected. 🚫

If set to 0 (the default), a large internally-defined limit will be applied.

If no unit is specified, bytes is assumed.

maxmessagesize: 0 — maximum size of messages that will be accepted by Cyrus. This affects LMTP deliveries, IMAP appends, DAV uploads, etc. Messages larger than this will be rejected. 📦

If set to 0 (the default), a large internally-defined limit will be applied.

If no unit is specified, bytes is assumed.

maxliteral: 128K — maximum size of a single literal allowed by the IMAP parser. 🔤

If set to 0, a large internally-defined limit will be applied.

If no unit is specified, bytes is assumed.

Literals used for message [part] data in APPEND are only limited by the 'maxmessagesize' option.

If the 'literalminus' option is enabled, non-synchonizing literals will be limited to the lesser of 4K and either 'maxliteral' or 'maxmessagesize', depending on the use-case.

maxquoted: 128K — maximum size of a single quoted string allowed by the IMAP parser. 💬

If set to 0, a large internally-defined limit will be applied.

If no unit is specified, bytes is assumed.

maxword: 128K — maximum size of a single word allowed by the IMAP parser. 📖

If set to 0, a large internally-defined limit will be applied.

If no unit is specified, bytes is assumed.

mboxkey_db: twoskip — the cyrusdb backend to use for mailbox keys. 🗄️

Allowed values: skiplist, twoskip, zeroskip

mboxlist_db: twoskip — the cyrusdb backend to use for the mailbox list. 🗂️

Allowed values: flat, skiplist, sql, twoskip, zeroskip

mboxlist_db_path — <none>

Absolute path to the mailboxes db file, bestie. Like, the whole path. If you don't specify it, it just defaults to configdirectory/mailboxes.db and honestly that's fine 💅

mboxname_lockpath — <none>

Path to the mailbox name lock files. The default is $conf/lock if you don't say anything 🎀

metapartition_files — <empty string>

A space-separated list of metadata files that get stored on a metapartition instead of in the mailbox directory on a spool partition. Allowed values are literally: header, index, cache, expunge, squat, annotations, lock, dav, archivecache ✨

metapartition-name — <none>

The pathname of the metadata partition name, matching up with the spool partition partition-name. So for any mailbox chilling in a directory on partition-name, the metadata files listed in metapartition_files get stored in a corresponding directory on metapartition-name. Not every partition-name option needs a matching metapartition-name option tho — so you can pick and choose which spool partitions get their own separate metadata partitions 🦋

mupdate_authname — <none>

The SASL username (Authentication Name) to use when authenticating to the mupdate server, if that's even needed 🔐

mupdate_config — standard

The configuration of the mupdate servers in the Cyrus Murder, and there's three flavors:

standard — there are discreet frontend (proxy) and backend servers
unified — a server can be both a frontend and a backend, main character energy fr
replicated — multiple backend servers all share the same mailspool, but each one has its own "replicated" copy of mailboxes.db

Allowed values: standard, unified, replicated 💖

munge8bit — 1

If this is enabled, lmtpd munges messages that have 8-bit characters in the headers. Those 8-bit characters get changed to X'. But heads up — if reject8bit is enabled, setting munge8bit does absolutely nothing 🙈 (A proper solution to non-ASCII characters in headers is offered by RFC 2047 and its predecessors, just saying.)

mupdate_connections_max — 128

The max number of connections a mupdate process will allow. This is tied to the number of file descriptors in the mupdate process. Go over this number and connections get immediately issued a BYE response. Rude but efficient 💀

mupdate_password — <none>

The SASL password (if needed) to use when authenticating to the mupdate server.

mupdate_port — 3905

The port of the mupdate server for the Cyrus Murder.

mupdate_realm — <none>

The SASL realm (if needed) to use when authenticating to the mupdate server.

mupdate_retry_delay — 20

The base time to wait between connection retries to the mupdate server. Patience is a virtue or whatever ⏳

mupdate_server — <none>

The mupdate server for the Cyrus Murder.

mupdate_username — <empty string>

The SASL username (Authorization Name) to use when authenticating to the mupdate server.

mupdate_workers_max — 50

The maximum number of mupdate worker threads, overall, periodt 👑

mupdate_workers_maxspare — 10

The maximum number of idle mupdate worker threads.

mupdate_workers_minspare — 2

The minimum number of idle mupdate worker threads.

mupdate_workers_start — 5

The number of mupdate worker threads to start. Like, at the beginning. The starter pack 🌸

netscapeurl — <none>

If enabled at compile time, this specifies a URL to reply when Netscape asks the server where the mail administration HTTP server is. Admins should set this to a local resource, no cap.

newsaddheaders — to

Space-separated list of headers to be added to incoming usenet articles. Added To: headers will contain email delivery addresses corresponding to each newsgroup in the Newsgroups: header. Added Reply-To: headers will contain email delivery addresses corresponding to each newsgroup in the Followup-To: or Newsgroups: header. And if the specified header(s) already exist in an article, the email delivery addresses get appended to the original header body(s) — no overwriting, we're just adding 💅

This option applies if and only if the newspostuser option is set. Allowed values: to, replyto

newsgroups — *

A wildmat pattern specifying which mailbox hierarchies should be treated as newsgroups. Only mailboxes matching the wildmat will accept and/or serve articles via NNTP. If you don't set it, a default wildmat of "*" (ALL shared mailboxes) gets used. And if the newsprefix option is also set, that default wildmat gets translated to "<newsprefix>.*" 🦋

newsmaster — news

The userid used for checking access controls when executing Usenet control messages. So like, to let articles be automatically deleted by cancel messages, give the "news" user the 'd' right on the desired mailboxes. To let newsgroups be automatically created, deleted and renamed by the corresponding control messages, give the "news" user the 'c' right on the desired mailbox hierarchies. Give them the keys to the kingdom 🔑

newspeer — <none>

A list of whitespace-separated news server specifications to which articles should be fed. Each server specification is a string of the form:

[user[:pass]@]host[:port][/wildmat]

where 'host' is the fully qualified hostname of the server, 'port' is the port the server is listening on, 'user' and 'pass' are the authentication credentials, and 'wildmat' is a pattern specifying which groups should be fed. If no 'port' is specified, port 119 is used. If no 'wildmat' is specified, all groups are fed. And if 'user' is specified (even if it's empty), then the NNTP POST command is used to feed the article to the server — otherwise the IHAVE command is used.

A '@' may be used in place of '!' in the wildmat to prevent feeding articles cross-posted to the given group. Otherwise, cross-posted articles are fed if any part of the wildmat matches. For example, the string "peer.example.com:*,!control.*,@local.*" would feed all groups except control messages and local groups to peer.example.com. And in the case of cross-posting to local groups, those articles would not be fed 🙅♀️

newspostuser — <none>

Userid used to deliver usenet articles to newsgroup folders (usually via lmtp2nntp). For example, if set to "post", email sent to "post+comp.mail.imap" would be delivered to the "comp.mail.imap" folder. Cute, right? 🎀

When set, the Cyrus NNTP server will add the header(s) specified in the newsaddheaders option to each incoming usenet article. The added header(s) will contain email delivery addresses corresponding to each relevant newsgroup. If not set, no headers are added to usenet articles.

newsprefix — <none>

Prefix to be prepended to newsgroup names to make the corresponding IMAP mailbox names.

newsrc_db_path — <none>

The absolute path to the newsrc db file. If not specified, it'll be configdirectory/fetchnews.db 💖

nntptimeout — 3m

Sets the length of the NNTP server's inactivity autologout timer. The minimum value is 3 minutes, which is also the default. So like, it times out fast and it's not sorry.

For backward compatibility, if no unit is specified, minutes is assumed. iykyk ⏰

notesmailbox — <none>

The top level mailbox in each user's account which is used to store Apple-style Notes. Default is blank (disabled). 🍎

notifysocket — {configdirectory}/socket/notify

Unix domain socket that the mail notification daemon listens on.

notify_external — <none>

Path to the external program that notifyd(8) will call to send mail notifications.

The external program will be called with the following command line options:

-c class
-p priority
-u user
-m mailbox

And the notification message will be available on stdin 📬

partition-name — <none>

The pathname of the partition name. At least one partition pathname MUST be specified. If the defaultpartition option is used, then its pathname MUST be specified too. For example, if the value of the defaultpartion option is part1, then the partition-part1 field is required. No skips, bestie 🚨

partition_select_mode — freespace-most

Partition selection mode. Here's the whole menu:

random — (pseudo-)random selection 🎲
freespace-most — partition with the most free space (KiB)
freespace-percent-most — partition with the most free space (%)
freespace-percent-weighted — each partition is weighted according to its free space (%); the more free space the partition has, the more chances it has to be selected
freespace-percent-weighted-delta — each partition is weighted according to its difference of free space (%) compared to the most used partition; the more the partition is lagging behind the most used partition, the more chances it has to be selected
- Note that actually even the most used partition has a few chances to be selected, and those chances increase when other partitions get closer 👀

Allowed values: random, freespace-most, freespace-percent-most, freespace-percent-weighted, freespace-percent-weighted-delta

partition_select_exclude — <none>

List of partitions to exclude from selection mode. Not invited to the party 💅

partition_select_usage_reinit — 0

For a given session, number of operations (e.g. partition selection) for which partitions usage data are cached.

partition_select_soft_usage_limit — 0

Limit of partition usage (%): if a partition is over that limit, it is automatically excluded from selection mode.

If all partitions are over that limit, this feature is not used anymore. Everyone's over the limit, so we're just ignoring the limit, okay? 😭

plaintextloginpause — <none>

Time to pause after a successful plaintext login. For systems that support strong authentication, this permits users to perceive a cost of using plaintext passwords. Petty? Yes. Effective? Also yes. (This does not affect the use of PLAIN in SASL authentications.)

For backward compatibility, if no unit is specified, seconds is assumed.

plaintextloginalert — <none>

Message to send to the client after a successful plaintext login. Say hi, be nice 💖

popexpiretime: -1 — okay so this is the duration that gets advertised as the minimum time a message might sit on the POP server before it's deleted (it's sent via the CAPA command, which is defined in the POP3 Extension Mechanism, and some clients actually support it). The granularity here is whole days, and partial days get truncated — so like "45m" is effectively "0d". The default is "NEVER", which you can specify with a negative number. 🚫🗓️

Now listen bestie: the Cyrus POP3 server never deletes mail, no matter what you set this to. But if your site has a less liberal policy, you gotta change this parameter accordingly. Also, for backward compatibility, if you don't specify a unit, days is assumed. 💅

popminpoll: <none> — this sets the minimum amount of time the server forces users to wait between successive POP logins. ⏳

For backward compatibility, if no unit is specified, minutes is assumed.

popsubfolders: 0 — allow access to subfolders of INBOX via POP3 by using the userid+subfolder syntax as the authentication/authorization id. 📂
poppollpadding: 1 — this creates a softer minimum poll restriction. It allows poppollpadding connections before the minpoll restriction is triggered. Additionally, one padding entry is recovered every popminpoll minutes. This allows for the occasional polling rate faster than popminpoll (i.e., for clients that require a send/receive to send mail) but still enforces the rate long-term. Default is 1 (disabled). 🛋️

The easiest way to think of it is a queue of past connections, with one slot being filled for every connection, and one slot being cleared every popminpoll minutes. When the queue is full, the user will not be able to check mail again until a slot is cleared. If the user waits a sufficient amount of time, they will get back many or all of the slots. 🎟️

poptimeout: 10m — set the length of the POP server's inactivity autologout timer. The minimum value is 10 minutes, the default. ⏰

For backward compatibility, if no unit is specified, minutes is assumed.

popuseacl: 0 — enforce IMAP ACLs in the pop server. Due to the nature of the POP3 protocol, the only rights which are used by the pop server are 'r', 't', and 's' for the owner of the mailbox. The 'r' right allows the user to open the mailbox and list/retrieve messages. The 't' right allows the user to delete messages. The 's' right allows messages retrieved by the user to have the \Seen flag set (only if popuseimapflags is also enabled). 🔐
popuseimapflags: 0 — if enabled, the pop server will set and obey IMAP flags. Messages having the \Deleted flag are ignored as if they do not exist. Messages that are retrieved by the client will have the \Seen flag set. All messages will have the \Recent flag unset. 🏷️
postmaster: postmaster — username that is used as the 'From' address in rejection MDNs produced by sieve. 📧
postuser: <empty string> — userid used to deliver messages to shared folders. For example, if set to "bb", email sent to "bb+shared.blah" would be delivered to the "shared.blah" folder. By default, an email address of "+shared.blah" would be used. 📬
proc_path: <none> — path to proc directory. Default is NULL — must be an absolute path if specified. If not specified, the path $configdirectory/proc/ will be used. 📁
prometheus_enabled: 0 — whether tracking of metrics for Prometheus is enabled. 📊
prometheus_need_auth: admin — authentication level required to fetch Prometheus metrics. 🔑

Allowed values: none, user, admin

prometheus_update_freq: 10s — deprecated in favour of prometheus_service_update_freq, prometheus_master_update_freq, and prometheus_usage_update_freq. ⚠️
prometheus_service_update_freq: 10s — frequency at which promstatsd should re-collate its service statistics report. The minimum value is 1 second. 🔄

If no unit is specified, seconds is assumed.

prometheus_master_update_freq: <none> — frequency at which master should re-collate its statistics report. If not set, prometheus_service_update_freq is used. 🔄

If no unit is specified, seconds is assumed.

prometheus_usage_update_freq: <none> — frequency at which promstatsd should re-collate its usage statistics report. Note that this report is relatively expensive to produce. If not set, usage statistics are not reported. 💸

Best to make this a multiple of prometheus_service_update_freq.

If no unit is specified, seconds is assumed.

prometheus_stats_dir: <none> — directory to use for gathering prometheus statistics. If specified, must be an absolute path. If not specified, the default path $configdirectory/stats/ will be used. It may be advantageous to locate this directory on ephemeral storage. 🗂️
proxy_authname: proxy — the authentication name to use when authenticating to a backend server in the Cyrus Murder. 🕵️
proxy_compress: 0 — try to enable protocol-specific compression when performing a client connection to a backend server in the Cyrus Murder. 🗜️

Note that this should only be necessary over slow network connections. Also note that currently only IMAP and MUPDATE support compression.

proxy_password: <none> — the default password to use when authenticating to a backend server in the Cyrus Murder. May be overridden on a host-specific basis using the hostname_password option. 🔑
proxy_realm: <none> — the authentication realm to use when authenticating to a backend server in the Cyrus Murder. 🌐
proxyd_allow_status_referral: 0 — set to true to allow proxyd to issue referrals to clients that support it when answering the STATUS command. This is disabled by default since some clients issue many STATUS commands in a row, and do not cache the connections that these referrals would cause, thus resulting in a higher authentication load on the respective backend server. 📨
proxyd_disable_mailbox_referrals: 0 — set to true to disable the use of mailbox-referrals on the proxy servers. 🚫
proxyservers: <none> — a list of users and groups that are allowed to proxy for other users, separated by spaces. Any user listed in this will be allowed to login for any other user: use with caution. In a standard murder this option should ONLY be set on backends. DO NOT SET on frontends or things won't work properly. ⚠️👑
pts_module: afskrb — the PTS module to use. 🧩

Allowed values: afskrb, ldap, http

ptloader_sock: <none> — Unix domain socket that ptloader listens on. (defaults to configdirectory/ptclient/ptsock). 🔌
ptscache_db: twoskip — the cyrusdb backend to use for the pts cache. 🗃️

Allowed values: skiplist, twoskip, zeroskip

ptscache_db_path: <none> — the absolute path to the ptscache db file. If not specified, will be configdirectory/ptscache.db 📍
ptscache_timeout: 3h — the timeout for the PTS cache database when using the auth_krb_pts authorization method (default: 3 hours). ⏳

For backward compatibility, if no unit is specified, seconds is assumed.

ptskrb5_convert524: 1 — when using the AFSKRB ptloader module with Kerberos 5 canonicalization, do the final 524 conversion to get an AFS style name (using '.' instead of '/', and using short names. 🔄
ptskrb5_strip_default_realm: 1 — when using the AFSKRB ptloader module with Kerberos 5 canonicalization, strip the default realm from the userid (this does not affect the stripping of realms specified by the afspts_localrealms option). ✂️
qosmarking: cs0 — this specifies the Class Selector or Differentiated Services Code Point designation on IP headers (in the ToS field). Allowed values: cs0, cs1, cs2, cs3, cs4, cs5, cs6, cs7, af11, af12, af13, af21, af22, af23, af31, af32, af33, af41, af42, af43, ef 🌐
quota_db: quotalegacy — the cyrusdb backend to use for quotas. 📊

Allowed values: flat, skiplist, sql, quotalegacy, twoskip, zeroskip

quota_db_path: <none> — the absolute path for the quota database (if you choose a single-file quota DB type — or the base path if you choose quotalegacy). If not specified will be configdirectory/quotas.db or configdirectory/quota/ 📁
quota_use_conversations: 0 — if conversations it enabled and quotaroot is a user folder, use the conversations quota counts, which count multiple copies of exactly the same message (by GUID) as only one. 💬
quotawarn: 90 — deprecated in favour of quotawarnpercent. ⚠️
quotawarnpercent: 90 — the percent of quota utilization over which the server generates warnings. ⚠️
quotawarnkb: <none> — deprecated in favour of quotawarnsize. ⚠️
quotawarnsize: 0 — the maximum amount of free space at which to give a quota warning (if this value is 0, or if the quota is smaller than this amount, then warnings are always given). 📏

Note that quota has kibibyte granularity. Values specified here will be truncated to the nearest whole kibibyte.

For backward compatibility, if no unit is specified, kibibytes is assumed.

quotawarnmsg: 0 — the maximum amount of messages at which to give a quota warning (if this value is 0, or if the quota is smaller than this amount, then warnings are always given). 📬
readonly: 0 — if enabled, all IMAP, POP and JMAP connections are read-only, no writes allowed. 🚫✍️
replicaonly: 0 — if enabled, marks the server as only being a replica. This will stop calalarmd from doing anything, and also deny any non-silent writes (nothing is allowed to increase modseq or uidvalidity). 🛑
reject8bit: 0 — if enabled, lmtpd rejects messages with 8-bit characters in the headers. 🚫🔤
restore_authname: <none> — the authentication used by the restore tool when authenticating to an IMAP/sync server. 🔄

`restore_password: <none>`

The password your restore tool uses when it's authenticating to an IMAP/sync server. Just the password, bestie, nothing else. 🔑

`restore_realm: <none>`

The authentication realm the restore tool uses when it's authenticating to an IMAP/sync server. It's giving "who are you, really" energy. 🎀

`reverseacls: 0`

When the server starts up, ctl_cyrusdb -r peeks at this value, and depending on what it says it'll either add or remove reverse ACL pointers from mailboxes.db. Yes, this one option literally changes what happens to your database at startup, so don't just vibes-based-edit it. 💅

`reverseuniqueids: 1`

Deprecated. No longer used. It's a ghost, bestie. Ignore it. 💀

`rfc2046_strict: 0`

Flip this on and imapd gets strict (like, per RFC 2046) when it's matching MIME boundary strings. Which means boundaries that contain other boundaries as substrings get treated as identical. Sounds tidy, right? Nope — enabling it will break some messages created by Eudora 5.1 (and earlier). So the recommendation is: leave it disabled unless you have a genuinely good reason to do otherwise. 🫶

`rfc2047_utf8: 0`

If enabled, imapd will parse any non-encoded character sequence in MIME header values as UTF8. Super useful for installations that either advertise the UTF8SMTP (RFC 5335) extension or receive mails with improperly escaped UTF-8 byte sequences. Same deal as above though — recommended that this option is left disabled unless there is good reason to do otherwise. ✨

`rfc3028_strict: 1`

If enabled, Sieve is strict (per RFC 3028) about which headers are allowed in address and envelope tests. Meaning: only those headers which are defined to contain addresses are allowed in address tests, and only "to" and "from" are allowed in envelope tests. When disabled, ANY grammatically correct header is allowed. Periodt. 🌈

`rss_feedlist_template: <none>`

A file containing HTML that gets used as a template for displaying the list of available RSS feeds. One single instance of the variable %RSS_FEEDLIST% should appear in the file, and that's where the nested unordered list of feeds gets swapped in. The toplevel unordered list gets tagged with an id of "feed" (<ul id='feed'>) which you can use with stylesheet(s) in your template. And the dynamically created list of feeds based on the HTML template will be accessible at the "/rss" URL on the server. Iconic. 🦋

`rss_feeds: *`

A wildmat pattern specifying which mailbox hierarchies get treated as RSS feeds. Only mailboxes matching the wildmat will have their messages available via RSS. If you don't set it, a default wildmat of "*" (that's ALL mailboxes) gets used. 👑

`rss_maxage: <none>`

Maximum age of items to display in an RSS channel. If non-zero, httpd will only display items received within this time period. If set to 0, all available items will be displayed (that's the default).

For backward compatibility, if no unit is specified, days is assumed. ⏳

`rss_maxitems: 0`

Maximum number of items to display in an RSS channel. If non-zero, httpd will display no more than the rss_maxitems most recent items. If set to 0, all available items will be displayed (the default). 🌸

`rss_maxsynopsis: 0`

Maximum RSS item synopsis length. If non-zero, httpd will display no more than the first rss_maxsynopsis characters of an item's synopsis. If set to 0, the entire synopsis will be displayed (the default). Short and sweet or the whole novel, your call. 📖

`rss_realm: <none>`

The realm to present for HTTP authentication of RSS feeds. If not set (which is the default), the value of the "servername" option will be used instead. 🎀

`sasl_auto_transition: 0`

If enabled, the SASL library will automatically create authentication secrets when given a plaintext password. See the SASL documentation for the deets — that's where the lore lives. 💜

`sasl_maximum_layer: 256`

Maximum SSF (security strength factor) that the server will allow a client to negotiate. The ceiling, basically. 🏛️

`sasl_minimum_layer: 0`

The minimum SSF that the server will allow a client to negotiate. A value of 1 requires integrity protection; any higher value requires some amount of encryption. So like, the floor. Don't go below the floor. 🔥

`sasl_option: 0`

Any SASL option can be set by preceding it with sasl_. And heads up — this file overrides the SASL configuration file. So what you write here wins. 💅

`sasl_pwcheck_method: <none>`

The mechanism used by the server to verify plaintext passwords. Possible values include "auxprop", "saslauthd", and "pwcheck". Pick your fighter. ⭐

`search_batchsize: 20`

The number of messages to be indexed in one batch (default 20). Note that long batches may delay user commands or mail delivery — so don't go feral with this one, bestie. 😭

`search_attachment_extractor_url: <none>`

A HTTP or HTTPS URL to extract search text from rich text attachments and other media during search indexing. The server at this URL must implement the following protocol, and I'm quoting it exactly because protocol is protocol:

For each attachment of an email, Cyrus sends a GET request to the URL <extractor-url>/<cyrus-id>, where <extractor-url> is the configured URL and <cyrus-id> is a Cyrus-chosen path segment that uniquely identifies this attachment.
If the extractor already has a cached plain text extract of the attachment identified by <cyrus-id> then it may return HTTP status code 200 (OK) and the plain text extract with a Content-Type "text/plain" header. Otherwise it must return HTTP status code 404 (Not Found).
If Cyrus receives the HTTP status code 404 (Not Found), then it sends a PUT request to the same URL as previously. The PUT request body contains the decoded, binary body of the attachment. The Content-Type request header has the same value as declared in the MIME part headers, including any type parameters.
The extractor must return the plain text extract with either HTTP status 200 (OK) or 201 (Created) and a Content-Type "text/plain" header. If no text can be extracted, then the extractor may return any return code in the range 4xx, or 200 and an empty response body.

Any other HTTP status code is treated as an error. For performance reasons, the Cyrus indexer attempts to keep-alive the TCP connection to the extractor. Xapian only. 🧚

`search_attachment_extractor_request_timeout: 5m`

Defines the duration after which to cancel non-responding requests to the search attachment extractor service.

If no unit is specified, seconds is assumed. ⏱️

`search_attachment_extractor_idle_timeout: 5m`

Defines the duration after which to close unused connections to the search attachment extractor service. If the idle timeout is less than search_attachment_extractor_request_timeout, then it is ignored and request timeout used instead.

If no unit is specified, seconds is assumed. 💤

`search_index_language: 0`

If enabled, then messages bodies are stemmed by detected language in addition to the default English stemmer. Xapian only. Multilingual slay. 🌍

`search_index_parts: 0`

Deprecated. No longer used. Another ghost. 💀

`search_index_skip_domains: <none>`

A space separated list of domains — if set, any users in the listed domains will be skipped when indexing. Not your problem, not your index. 🙈

`search_index_skip_users: <none>`

A space separated list of usernames — if set, any users in the list will be skipped when indexing. Same energy, different list. 🙈

`search_query_language: 0`

Deprecated. No longer used. Bye. 👋

`search_normalisation_max: 1000`

A resource bound for the combinatorial explosion of search expression tree complexity caused by normalising expressions with many OR nodes. Those can use more CPU time to optimise than they save IO time in scanning folders. So it's a vibe check on your OR spam. 🔥

`search_engine: none`

The indexing engine used to speed up searching.

Allowed values: none, squat, xapian 👑

`search_fuzzy_always: 0`

Whether to enable RFC 6203 FUZZY search for all IMAP SEARCH. If turned on, search attributes will be searched using FUZZY search by default. If turned off, clients have to explicitly use the FUZZY search key to enable fuzzy search for regular SEARCH commands. 🦋

`search_index_headers: 1`

Whether to index headers other than From, To, Cc, Bcc, and Subject. Experiment shows that some headers such as Received and DKIM-Signature can contribute up to 2/3rds of the index size but almost nothing to the utility of searching. Note that if header indexing is disabled, headers can still be searched, the searches will just be slower. 🌸

`search_indexed_db: twoskip`

The cyrusdb backend to use for the search latest indexed uid state. Xapian only.

Allowed values: flat, skiplist, twoskip, zeroskip 💅

`search_maxtime: <none>`

The maximum number of seconds to run a search for before aborting. Default of no value means search "forever" until other timeouts. Forever search era. ♾️

`search_maxsize: 4M`

The maximum size to index for each message part. Message contents that occur after this byte offset will not be indexed nor used to generate search snippets. Xapian-only.

For backward compatibility, if no unit is specified, kibibytes is assumed. 📏

`search_queryscan: 5000`

The minimum number of records require to do a direct scan of all G keys * rather than indexed lookups. A value of 0 means always do indexed lookups. (Yes, the original says "require" — I'm not fixing it, I'm just reporting the tea. 🍵)

`search_skipdiacrit: 1`

When searching, should diacriticals be stripped from the search terms. The default is "true", so a search for "hav" will match "Håvard". This is not RFC 5051 compliant, but is backwards compatible, and may be preferred by some sites. ⭐

`search_skiphtml: 0`

If enabled, HTML parts of messages are skipped, i.e. not indexed and not searchable. Otherwise, they're indexed. 🎀

`search_whitespace: merge`

When searching, how whitespace should be handled. The options are:

"skip" (default in 2.3 and earlier series) — where a search for "equi" would match "the quick brown fox".
"merge" — the default, where "he qu" would match "the quick brownfox".
"keep" — where whitespace must match exactly.

The default of "merge" is recommended for most cases — it's a good compromise which keeps words separate.

Allowed values: skip, merge, keep ✨

search_snippet_length: 255 — the max byte length of a snippet that the XSNIPPETS command spits out. Only the Xapian search backend supports this, and bestie, it will try to always fill all search_snippet_length bytes of that snippet. Overachiever energy fr. ✨
search_stopword_path: <none> — the absolute base path to the search stopword lists. Leave it unset and literally no stopwords get taken into account during search indexing, which is giving "no filter" but for search. Currently the only supported and default stop word file is english.txt.
searchpartition-name: <none> — the pathname where you store the Xapian search indexes of searchtier for mailboxes of partition name. You MUST configure this for the defaultsearchtier and for any additional search tier (peep squatter for the deets).

Example, okay, because this one's a lot: if defaultpartition is defined as part1 and defaultsearchtier as tier1, then your config must contain an entry tier1searchpartition-part1 that defines the path where to store this tier1's search index for the part1 partition.

This option MUST be specified for Xapian search. No cap. 🫶

seenstate_db: twoskip — the cyrusdb backend to use for the seen state.

Allowed values: flat, skiplist, twoskip, zeroskip

sendmail: /usr/lib/sendmail — the pathname of the sendmail executable. Sieve invokes sendmail for sending rejections, redirects and vacation responses. 💌
sendmail_auth_id: CYRUS_SENDMAIL_AUTH_ID — the name of an environment variable to set when invoking sendmail. That environment variable's value will contain the user id of the currently authenticated user. If nobody's authenticated, the env var just... doesn't get set. 🙈
serverlist: <none> — a whitespace separated list of backend server names. Used for finding the server with the most available free space for proxying CREATE. Basically it's your server roster. 👑
serverlist_select_mode: freespace-most — the server selection mode, i.e. how we pick which backend gets the job.
- random — (pseudo-)random selection
- freespace-most — backend with the most (total) free space (KiB)
- freespace-percent-most — backend whose partition has the most free space (%)
- freespace-percent-weighted — same as for partition selection, comparing the free space (%) of the least used partition of each backend
- freespace-percent-weighted-delta — same as for partition selection, comparing the free space (%) of the least used partition of each backend.

Allowed values: random, freespace-most, freespace-percent-most, freespace-percent-weighted, freespace-percent-weighted-delta

serverlist_select_usage_reinit: 0 — for a given session, the number of operations (e.g. backend selection) for which backend usage data are cached. ✨
serverlist_select_soft_usage_limit: 0 — a limit of backend usage (%). If a backend is over that limit, it's automatically excluded from selection mode. And if all backends are over that limit? Then this feature just isn't used anymore. Everybody's over the limit, so we pretend the limit doesn't exist. 💀
servername: <none> — the hostname visible in the greeting messages of the POP, IMAP and LMTP daemons. Unset it and whatever gethostname(2) returns gets used instead. This is also the value murder clusters use to identify the host name. It should be resolvable by DNS to the correct host, and unique within an active cluster. And if you're using low level replication (like drbd) then it should be the same on each copy and the DNS name should also be moved to the new master on failover. Cluster girlies, write this one down.
serverinfo: on — the server information to display in the greeting and capability responses. Info shows up like this:
- "off" = no server information in the greeting or capabilities
- "min" = servername in the greeting; no server information in the capabilities
- "on" = servername and product version in the greeting; product version in the capabilities

Allowed values: off, min, on

sharedprefix: Shared Folders — if you're using the alternate IMAP namespace, this is the prefix for the shared namespace. The hierarchy delimiter gets automatically appended. No manual delimiter-ing, we love a self-sufficient option. 💅
sieve_allowreferrals: 1 — if enabled, timsieved will issue referrals to clients when the user's scripts live on a remote server (in a Murder). Otherwise, timsieved will proxy traffic to the remote server. Pick your fighter. 🥊
sieve_duplicate_max_expiration: 90d — maximum expiration time for duplicate message tracking records.

For backward compatibility, if no unit is specified, seconds is assumed.

sieve_extensions: fileinto reject vacation vacation-seconds notify include envelope environment body relational regex subaddress copy date index imap4flags mailbox mboxmetadata servermetadata variables editheader extlists duplicate ihave fcc special-use redirect-dsn redirect-deliverby mailboxid vnd.cyrus.log vnd.cyrus.jmapquery processcalendar snooze vnd.cyrus.implicit_keep_target — a space-separated list of Sieve extensions allowed to be used in sieve scripts, enforced at submission by timsieved(8). Any previously installed script will be unaffected by this option and will continue to execute regardless of the extensions used. This option has no effect on options that are disabled at compile time (e.g., "regex"). 🎀

Allowed values: fileinto, reject, vacation, vacation-seconds, notify, include, envelope, environment, body, relational, regex, subaddress, copy, date, index, imap4flags=imapflags, mailbox, mboxmetadata, servermetadata, variables, editheader, extlists, duplicate, ihave, fcc, special-use, redirect-dsn, redirect-deliverby, mailboxid, vnd.cyrus.log=x-cyrus-log, vnd.cyrus.jmapquery=x-cyrus-jmapquery, processcalendar=vnd.cyrus.imip, snooze=vnd.cyrus.snooze=x-cyrus-snooze, vnd.cyrus.implicit_keep_target

sieve_folder: #sieve — the name of the folder for storing Sieve scripts. 📁
sieve_maxscriptsize: 32K — maximum size any sieve script can be, enforced at submission by timsieved(8) and JMAP.

For backward compatibility, if no unit is specified, kibibytes is assumed.

sieve_maxscripts: 5 — maximum number of sieve scripts any user may have, enforced at submission by timsieved(8). Five is the moment. ✋
sieve_utf8fileinto: 0 — if enabled, the sieve engine expects folder names for the fileinto action in scripts to use UTF8 encoding. Otherwise, modified UTF7 encoding should be used. Encoding lore, iykyk.
sieve_sasl_send_unsolicited_capability: 1 — Deprecated: this option does nothing. It's giving decorative. Ignore her. 💀
sieve_use_lmtp_reject: 1 — enabled by default. If reject can be done via LMTP, then return a 550 rather than generating the bounce message in Cyrus. Straight to the point, no bounce middleman. 🔥
sieve_vacation_min_response: 3d — minimum time interval between consecutive vacation responses, per RFC 6131. The default is 3 days.

For backward compatibility, if no unit is specified, seconds is assumed.

sieve_vacation_max_response: 90d — maximum time interval between consecutive vacation responses, per RFC 6131. The default is 90 days. The minimum is 7 days.

For backward compatibility, if no unit is specified, seconds is assumed.

sievedir: /usr/sieve — if sieveusehomedir is false, this directory is searched for Sieve scripts. 🦋
sievenotifier: <none> — the Notifyd(8) method to use for "SIEVE" notifications. If it's not set, "SIEVE" notifications are disabled. This method is only used when no method is specified in the script.
sieveusehomedir: 0 — if enabled, lmtpd will look for Sieve scripts in user's home directories: ~user/.sieve. 🏠
anysievefolder: 0 — it must be "yes" in order to permit the autocreation of any INBOX subfolder requested by a sieve filter, through the "fileinto" action.
singleinstancestore: 1 — if enabled, imapd, lmtpd and nntpd attempt to only write one copy of a message per partition and create hard links, resulting in a potentially large disk savings. Disk savings?? Obsessed. 💖
skiplist_always_checkpoint: 1 — if enabled, this option forces the skiplist cyrusdb backend to always checkpoint when doing a recovery. This causes slightly more IO, but on the other hand leads to more efficient databases, and the entire file is already "hot". Trade-offs, bestie. ✨
skiplist_unsafe: 0 — if enabled, this option forces the skiplist cyrusdb backend to not sync writes to the disk.

Enabling this option is NOT RECOMMENDED. Like, at all. Do not. 🚨

smtp_backend: sendmail — the SMTP backend to use for sending email. 📧

The "host" backend sends message submissions via a TCP socket to the SMTP host defined in the config option smtp_host.

The "sendmail" backend forks the Cyrus process into the executable defined in the config option sendmail. The executable must accept "-bs" as command line argument, read from stdin and must implement the minimum SMTP protocol as defined in section 4.5.1 of RFC 5321.

If the SMTP EHLO command reports AUTH (RFC 4954) as a supported extension, then the MAIL FROM command includes the AUTH parameter, with its value set to the name of any authenticated user which triggered the email. The AUTH parameter is omitted if the user is unknown to the calling process.

If the directory configdirectory/log/smtpclient.smtp_backend exists, then telemetry logs for outgoing SMTP sessions will be created in this directory.

Allowed values: host, sendmail

smtp_host: localhost:587 — the SMTP host to use for sending mail (also see the smtp_backend option). The value of this option must be the name or IP address of a TCP host, followed optionally by a colon and the port or service to use. The default port is 587. TLS may be activated by appending "/tls" to the value. Authentication is enabled if smtp_auth_authname is set. Authentication can be explicitly disabled by appending "/noauth" to the host address. 🌈
smtp_auth_authname: <none> — the authentication name to use when authenticating to the SMTP server defined in smtp_host. 🔑
smtp_auth_password — default: <none>. The password to use when authenticating to the SMTP server defined in smtp_host. Keep it secret, keep it safe 💖
smtp_auth_realm — default: <none>. The authentication SASL realm to use when authenticating to a SMTP server. SASL realm realness ✨
soft_noauth — default: 1. If enabled, lmtpd returns temporary failures if the client does not successfully authenticate. Otherwise lmtpd returns permanent failures (causing the mail to bounce immediately). So if it's on, it's a soft no; if it's off, the mail bounces instantly 💀
sortcache_db — default: twoskip. The cyrusdb backend to use for caching sort results (currently only used for xconvmultisort). Allowed values: skiplist, twoskip, zeroskip. Sorting cache vibes 🧚
specialuse_extra — default: <none>. Whitespace separated list of extra special-use attributes that can be set on a mailbox. RFC 6154 currently lists what special-use attributes can be set. This allows extending that list in the future or adding your own if needed. Extra special-use, because why not 👑
specialuse_nochildren — default: \Scheduled \Snooze. Whitespace separated list of special-use attributes that may not contain child folders. If set, mailboxes with any of these attributes may not have child folders created, and these attributes cannot be added to mailboxes that already have children. No kids allowed under these folders 🚫
specialuse_protect — default: \Archive \Drafts \Important \Junk \Sent \Scheduled \Snooze \Trash. Whitespace separated list of special-use attributes to protect the mailboxes for. If set, don't allow mailboxes with these special use attributes to be deleted or renamed to have a different parent. Default is the built-in list. Protect these at all costs 🛡️
specialusealways — default: 1. If enabled, this option causes unextended LIST and LSUB output to include the "special-use" flags. Always show the special-use flags 💅
sql_database — default: <none>. Name of the database which contains the cyrusdb table(s). SQL database name drop 💖
sql_engine — default: <none>. Name of the SQL engine to use. Allowed values: mysql, pgsql, sqlite. Pick your SQL engine, bestie ✨
sql_hostnames — default: <empty string>. Comma separated list of SQL servers (in host[:port] format). Host:port combo list, no cap 🌸
sql_passwd — default: <none>. Password to use for authentication to the SQL server. SQL password, keep it lowkey 🔐
sql_user — default: <none>. Username to use for authentication to the SQL server. SQL username, main character energy 👑
sql_usessl — default: 0. If enabled, a secure connection will be made to the SQL server. Secure the bag with SSL 💚
srs_alwaysrewrite — default: 0. If true, perform SRS rewriting for ALL forwarding, even when not required. Rewrite ALL the things 🔥
srs_domain — default: <none>. The domain to use in rewritten addresses. This must point only to machines which know the encoding secret used by this system. When present, SRS is enabled. Only point to machines that know the secret 🤫
srs_hashlength — default: 0. The hash length to generate in a rewritten address. Hash length, cute ✨
srs_secrets — default: <none>. A list of secrets with which to generate addresses. Secrets list, shh 🤫
srs_separator — default: <none>. The separator to appear immediately after SRS[01] in rewritten addresses. Separator after SRS[01] 💅
srvtab — default: <empty string>. Deprecated: This option does nothing. Deprecated, it does nothing. Not me wasting time 💀
submitservers — default: <none>. A list of users and groups that are allowed to resolve "urlauth=submit+" IMAP URLs, separated by spaces. Any user listed in this will be allowed to fetch the contents of any valid "urlauth=submit+" IMAP URL: use with caution. Use with caution, bestie! This is powerful 🚨
subscription_db — default: flat. The cyrusdb backend to use for the subscriptions list. Allowed values: flat, skiplist, twoskip, zeroskip. Subscription list backend vibes 🧚
suppress_capabilities — default: <none>. Suppress the named capabilities from any capability response. Use the exact case as it appears in the response, e.g. "suppress_capabilities: ESEARCH QRESYNC WITHIN XLIST LIST-EXTENDED" if you have a murder with 2.3.x backends and don't want clients being confused by new capabilities that some backends don't support. Exact case, no cap! 👑
statuscache — default: 0. Enable/disable the imap status cache. Status cache on/off 💅
statuscache_db — default: twoskip. The cyrusdb backend to use for the imap status cache. Allowed values: skiplist, sql, twoskip, zeroskip. Status cache backend, slay ✨
statuscache_db_path — default: <none>. The absolute path to the statuscache db file. If not specified, will be configdirectory/statuscache.db. Absolute path or default to configdirectory/statuscache.db 🌈
sync_authname — default: <none>. The authentication name to use when authenticating to a sync server. Prefix with a channel name to only apply for that channel. Sync auth name, channel-prefixed 💖
sync_batchsize — default: 8192. the number of messages to upload in a single mailbox replication. Default is 8192. If there are more than this many messages appended to the mailbox, generate a synthetic partial state and send that. Batch size 8192, synthetic partial state if more 🦋
sync_cache_db — default: twoskip. The cyrusdb backend to use for the replication cache. Allowed values: skiplist, sql, twoskip, zeroskip. Replication cache backend ✨
sync_cache_db_path — default: <none>. The path for the replication cache. Prefix with a channel name to apply for that channel. NOTE, it's quite important to have a different one per backend! NOTE: different one per backend, fr fr! 🔥
sync_host — default: <none>. Name of the host (replica running sync_server(8)) to which replication actions will be sent by sync_client(8). Prefix with a channel name to only apply for that channel. Sync host, where the replica lives 👑
sync_log — default: 0. Enable replication action logging by lmtpd(8), imapd(8), pop3d(8), and nntpd(8). The log {configdirectory}/sync/log is used by sync_client(8) for "rolling" replication. Rolling replication logging 💅
sync_log_chain — default: 0. Enable replication action logging by sync_server as well, allowing chaining of replicas. Use this on 'B' for A => B => C replication layout. Chain replicas A => B => C 🔗
sync_log_channels — default: <none>. If specified, log all events to multiple log files in directories specified by each "channel". Each channel can then be processed separately, such as by multiple sync_client(8)s in a mesh replication scheme, or by squatter(8) for rolling search index updates. You can use "" (the two-character string U+22 U+22) to mean the default sync channel. Multi-channel logging, mesh vibes 🌈
sync_log_unsuppressable_channels — default: squatter. If specified, the named channels are exempt from the effect of setting sync_log_chain:off, i.e. they are always logged to by the sync_server process. This is only really useful to allow rolling search indexing on a replica. Always logged, even if chain is off ✨
sync_password — default: <none>. The default password to use when authenticating to a sync server. Prefix with a channel name to only apply for that channel. Sync password, channel-prefixed 💖
sync_port — default: <none>. Name of the service (or port number) of the replication service on replica host. Prefix with a channel name to only apply for that channel. If not specified, and if sync_try_imap is set to "yes" (the default), then the replication client will first try "imap" (port 143) to check if imapd supports replication. Otherwise it will default to "csync" (usually port 2005). Port or service, tries imap 143 then csync 2005 🦋
sync_realm — default: <none>. The authentication realm to use when authenticating to a sync server. Prefix with a channel name to only apply for that channel. Sync realm, channel-prefixed 👑
sync_reconnect_maxwait — default: 20m. When sync_client cannot connect to the replica, it enters a retry loop with an exponential backoff between attempts. This option sets the upper limit on that exponential backoff: no matter how long the replica has been down so far, sync_client will never wait longer than sync_reconnect_maxwait between retries. If this is zero or negative, the backoff duration will be allowed to increase indefinitely (not recommended). If no unit is specified, seconds is assumed. Not recommended to go infinite, bestie! ⏳
sync_repeat_interval — default: 1s. Minimum interval between replication runs in rolling replication mode. If a replication run takes longer than this time, we repeat immediately. Prefix with a channel name to only apply for that channel. For backward compatibility, if no unit is specified, seconds is assumed. Rolling mode repeat interval 💅
sync_rightnow_channel — default: <none>. if set, run sync_client to this channel immediately. As with channels, set this value to '""' to sync the default channel! Sync right now! Set to '""' for default 🚀
sync_shutdown_file — default: <none>. Simple latch used to tell sync_client(8) that it should shut down at the next opportunity. Safer than sending signals to running processes. Prefix with a channel name to only apply for that channel. Safer than signals, fr fr 🛑
sync_timeout — default: 30m. How long to wait for a response before returning a timeout failure when talking to a replication peer (client or server). The minimum duration is 3 seconds, the default is 30 minutes. For backward compatibility, if no unit is specified, seconds is assumed. Timeout, min 3 seconds, default 30 minutes ⏰
sync_try_imap — default: 1. Whether sync_client should try to perform an IMAP connection before falling back to csync. If this is set to "no", sync_client will only use csync. Prefix with a channel name to apply only for that channel. Try IMAP first, then csync ✨
syslog_prefix — default: <none>. String to be prepended to the process name in syslog entries. Can be further overridden by setting the $CYRUS_SYSLOG_PREFIX environment variable. Using the $CYRUS_SYSLOG_PREFIX environment variable has the additional advantage that it can be set before the imapd.conf is read, so errors while reading the config file can be syslogged with the correct prefix. Prefix perks: can be set before config read 💖
syslog_facility — default: <none>. Configure a syslog facility. The default is whatever is compiled in. Allowed values are: DAEMON, MAIL, NEWS, USER, and LOCAL0 through to LOCAL7. Syslog facility, default is compiled in 🌸
tcp_keepalive — default: 0. Enable keepalive on TCP connections. TCP keepalive on/off 💅
tcp_keepalive_cnt — default: 0. Number of TCP keepalive probes to send before declaring the connection dead (0 == system default). Probe count, 0 means system default 👑
tcp_keepalive_idle — default: 0. How long a connection must be idle before keepalive probes are sent (0 == system default). For backward compatibility, if no unit is specified, seconds is assumed. Idle time, 0 means system default ⏳
tcp_keepalive_intvl — default: 0. Time between keepalive probes (0 == system default). For backward compatibility, if no unit is specified, seconds is assumed. Interval between probes, 0 means system default ✨
temp_path — default: /tmp. The pathname to store temporary files in. It is recommended to use an in-memory filesystem such as tmpfs for this path. Recommended: tmpfs for temp_path 🧚
telemetry_bysessionid — default: 0. If true, log by sessionid instead of PID for telemetry. Log by sessionid, not PID 💅
timeout: 32m — how long the IMAP server waits around being ignored before it kicks you out (the inactivity autologout timer, bestie). Minimum you're allowed to set is 30 minutes. Default is 32 minutes, and that lil extra cushion exists on purpose so clients that NOOP every 30 minutes don't get got. 🫶

For backward compatibility, if you don't specify a unit, minutes is assumed.

imapidletimeout: <none> — timeout specifically for idling clients (RFC 2177, the IDLE girlies). If it's not set (which is the default), the value of "timeout" gets used instead. 💤

For backward compatibility, if no unit is specified, minutes is assumed.

tls_ca_file: <none> — deprecated, babe. Use tls_client_ca_file instead.
tls_ca_path: <none> — also deprecated, use tls_client_ca_dir instead.
tlscache_db: twoskip — deprecated in favor of tls_sessions_db. 👋
tlscache_db_path: <none> — deprecated in favor of tls_sessions_db_path.
tls_cert_file: <none> — deprecated in favor of tls_server_cert.
tls_cipher_list: DEFAULT — deprecated in favor of tls_ciphers.
tls_ciphers: DEFAULT — the list of SSL/TLS ciphers you're allowing. The format of that string (and what "DEFAULT" even means) is described in ciphers(1). 🔐

Also check Mozilla's server-side TLS recommendations, they're iconic:

https://wiki.mozilla.org/Security/Server_Side_TLS

tls_crl_file: <none> — path to a file holding the Certificate Revocation List.
tls_client_ca_dir: <none> — path to a directory holding the CA certificates used to verify client SSL certificates used for authentication.
tls_client_ca_file: <none> — path to a file holding the CA certificate(s) used to verify client SSL certificates used for authentication.
tls_client_cert: <none> — file containing the certificate presented to a server for authentication during STARTTLS. Setting it to "disabled" turns off this server's use of certificate-based authentication. 🙈
tls_client_certs: optional — disable ("off"), allow ("optional", the default) or require ("require") clients using SSL certificates to authenticate themselves. Allowed values: off, optional, require.
tls_client_key: <none> — file containing the private key that belongs to the tls_client_cert certificate. A value of "disabled" disables this server's use of certificate-based authentication.
tls_eccurve: prime256v1 — the elliptic curve used for ECDHE. Default is NIST Suite B prime256. Run 'openssl ecparam -list_curves' to see all the possible values, go look. 🌀
tls_key_file: <none> — deprecated in favor of tls_server_key.
tls_required: 0 — if you enable this, a TLS/SSL encryption layer has to be negotiated before ANY authentication mechanisms get advertised or allowed. No encryption, no auth, periodt. 🔥
tls_prefer_server_ciphers: 0 — prefer the ciphers on the server side instead of the client side.
tls_server_ca_dir: <none> — path to a directory with CA certificates used to verify certificates offered by the server, for when cyrus is acting as a client. This directory must have filenames with the hashed value of the certificates (peep openssl(1)).
tls_server_ca_file: <none> — path to a file containing CA certificates used to verify certificates offered by the server, when cyrus acts as a client.
tls_server_cert: <none> — file containing the certificate, including the full chain, presented to clients. You can set two certificates, like RSA and EC, if the filenames are separated with a comma and no spaces.
tls_server_dhparam: <none> — file containing the DH parameters belonging to the certificate in tls_server_cert. Used by OpenSSL before version 3.0.
tls_server_key: <none> — file containing the private key belonging to the certificate in tls_server_cert. If it's not set, tls_server_cert must contain both the private and public key. If you're using two certificates you can set two key files, and then the files must be separated with a comma and no spaces.
tls_sessions_db: twoskip — the cyrusdb backend to use for the TLS cache. 🗂️

Allowed values: skiplist, sql, twoskip, zeroskip

tls_sessions_db_path: <none> — the absolute path to the TLS sessions db file. If you don't specify it, it'll be configdirectory/tls_sessions.db.
tls_session_timeout: 24h — how long a TLS session gets cached for later reuse. The maximum value is 24 hours, which is also the default. A value of 0 disables session caching entirely. ⏰

For backward compatibility, if no unit is specified, minutes is assumed.

tls_versions: tls1_0 tls1_1 tls1_2 tls1_3 — a list of SSL/TLS versions to not disable. Cyrus IMAP SSL/TLS starts with all protocols and then subtracts the ones not in this list. Newer versions of SSL/TLS will need to be added here if you want them to be able to get disabled. 💅
uidl_format: cyrus — pick the format for UIDLs in pop3. Possible values are "uidonly", "cyrus", "dovecot" and "courier". "uidonly" forces the old default of UID, "cyrus" is UIDVALIDITY.UID. Dovecot is 8 digits of leading hex (lower case) each UID UIDVALIDITY. Courier is UIDVALIDITY-UID. Allowed values: uidonly, cyrus, dovecot, courier
umask: 077 — the umask value used by various Cyrus IMAP programs.
userdeny_db: flat — the cyrusdb backend to use for the user access list.

Allowed values: flat, skiplist, sql, twoskip, zeroskip

userdeny_db_path: <none> — the absolute path to the userdeny db file. If you don't specify it, it'll be configdirectory/user_deny.db.
username_tolower: 1 — convert usernames to all lowercase before login/authentication. Super useful with authentication backends that ignore case during username lookups (hi, LDAP 👋).
userprefix: Other Users — if you're using the alternate IMAP namespace, this is the prefix for the other users namespace. The hierarchy delimiter gets automatically appended.
unix_group_enable: 1 — should we look up groups when using auth_unix? Disable this if you're not using groups in ACLs for your IMAP server and you're using auth_unix with a backend (like LDAP) that can make getgrent() calls painfully slow. 🐌
unixhierarchysep: 1 — use the UNIX separator character '/' for delimiting levels of mailbox hierarchy. Turn it off to use the netnews separator character '.'. Heads up though: with the newnews separator, no dots may occur in mailbox names. The default switched in 3.0 from off to on.
vcard_max_size: 0 — maximum allowed vCard size. If it's non-zero, CardDAV and JMAP will reject storage of contacts whose vCard representation is bigger than vcard_max_size. 📇

If set to 0 (the default), a large internally-defined limit gets applied.

If no unit is specified, bytes is assumed.

virtdomains: off — configure virtual domain support. Here are the moods:
- off — Cyrus doesn't know or care about domains. Only the local part of email addresses is ever considered. This is not recommended for any deployment, but it's currently the default.
- userid — the user's domain is determined by splitting a fully qualified userid at the last '@' or '%' symbol. If the userid is unqualified, the defaultdomain gets used. This is the recommended configuration for all deployments. If you wish to provide calendaring services you must use this configuration.
- on — fully qualified userids are respected, same as "userid". Unqualified userids will have their domain determined by doing a reverse lookup on the IP address of the incoming network interface, or if no record is found, the defaultdomain gets used.

Allowed values: off, userid, on

virusscan_notification_subject: Automatically deleted mail — the text used in the subject of email notifications created by cyr_virusscan(8) when it deletes infected mail. 🦠
virusscan_notification_template: <none> — the absolute path to a file containing a template to use to describe infected messages that have been deleted by cyr_virusscan(8). See cyr_virusscan(8) for the spec of how this file's format works. If you don't specify it, the builtin default template gets used.
websocket_timeout: 30m — sets the length of the HTTP server's inactivity autologout timer once a WebSocket channel has been established. Default is 30 minutes. The minimum value is 0, which disables WebSockets. 🔌

If no unit is specified, minutes is assumed.

webdav_attachments_baseurl: <none> — the base URL for WebDAV managed attachments, not counting the Cyrus-specific URL paths. Typically this only includes the HTTP scheme and host name at which DAV clients may fetch managed attachments from, e.g. "https://dav.example.com". If it's not specified, the target host of the DAV HTTP request that uploaded the attachment gets used. For JMAP, the Link.blobId property of JSCalendar objects will be disabled.
webdav_attachments_max_binary_attach_size: 1024K — the maximum byte length of an ATTACH property value when managed attachment URIs get rewritten to BINARY during iTIP processing. Any attachment that exceeds this byte size keeps its managed attachment URI as its ATTACH value. A zero size limit causes attachments of any size to be rewritten as BINARY.
xbackup_enabled: 0 — deprecated. No longer used, she's retired. 💀
xlist-flag: <none> — set the special-use flag flag on the specified folder when it gets autocreated (see the autocreate_inbox_folders option). For example, if xlist-junk: Spam is set, and the folder Spam is autocreated, the special-use flag \Junk gets set on it. 🗑️

(This option is so named for backward compatibility with old config files. iykyk.)

lmtp_catchall_mailbox: <none> — mail sent to mailboxes that don't exist will be delivered to this user.

NOTE: This must be an existing local user name with an INBOX, NOT an email address! Don't get caught slippin' on this one. 😭

zoneinfo_db: twoskip — the cyrusdb backend for zoneinfo. This db is what the tzdist httpmodules use, and it's managed by ctl_zoneinfo(8). Allowed values, and bestie you have exactly four: flat, skiplist, twoskip, zeroskip. That's the whole menu. 💅
zoneinfo_db_path: <none> — the absolute path to the zoneinfo db file. Leave it blank and it just becomes configdirectory/zoneinfo.db. Easy. ✨
zoneinfo_dir: <none> — the absolute path to the zoneinfo directory, the one holding all your timezone definitions as generated by the vzic tool. Setting this overrides any compile-time value that got baked in with the --with-zoneinfo-dir configure option. Rude? No, it's just how precedence works. 🎀

If you don't specify it, and there's no compile-time value either, then whatever definitions libical manages to find are what you get. Surprise time. 🦋

And if you're running a Time Zone Data Distribution Service — like, you've literally got "tzdist" sitting in your httpmodules — then this option or a compile-time value MUST be specified. Not optional. Not vibes-based. Required, periodt.

object_storage_enabled: 0 — is Object storage enabled for this server? Big question. Note that you also need archiving enabled and an archivepartition for the mailbox — this flag alone doesn't do the thing. Only email files go on Object storage; the archive partition is what stores any other files. Teamwork. 🫶
object_storage_dummy_spool: <none> — a dummy object storage spool. This one is for test only, okay? It's the spool where the user directory (container) gets created to store all the emails in a flat structure. 🙈
openio_namespace: <none> — the OpenIO namespace used to store archived email messages. A namespace identifies the physical platform that cyrus has to contact. This directive is what the OpenIO's SDK uses to locate its platform entry point. 📍
openio_account: <none> — the OpenIO account used to account for stored emails. Accounts are unique inside their namespace, and they hand you virtual partitions, with quotas and QoS features. Fancy. 👑
openio_rawx_timeout: 30s — the OpenIO timeout for querying the RAWX services (default 30 sec). ⏳
openio_proxy_timeout: 5s — the OpenIO timeout for querying the PROXY services (default 5 sec). ⏳
openio_autocreate: 0 — lets the OpenIO SDK autocreate containers. This is mainly destined to be turned on in development environments. In production, the container should have already been provisioned with the mailboxes, so like… don't. 💀
openio_verbosity: <none> — sets the logging verbosity of the OpenIO's internal behavior. Admissible values are "warning", "notice", "info", "debug", "trace", "quiet". The default verbosity is "warning". Set to "notice" for a few lines on a per-client basis. Set to "info" for a few lines on a per-request basis. Set to "debug". Set to "trace" to activate the underlying libcurl debug output. Enabling a verbosity higher to equal than "debug" requires the cyrus to be set in debug mode. And the special "quiet" value disables all kinds of logging at the GLib level. 🌈
caringo_hostname: <none> — the Caringo hostname used to store archived email messages. A hostname identifies the physical platform cyrus must contact. This directive is used by the Caringo's SDK (CastorSDK: Caringo Simple Content Storage Protocol (SCSP) on HTTP 1.1 using a RESTful architecture. 🍓
caringo_port: 80 — the port of the caringo server (caringo_hostname); default is 80. 🔌
fastmailsharing: 0 — if enabled, use FastMail style sharing (oldschool full server paths). Vintage. Iconic. ✨

Go read them, bestie, they're all part of the same family reunion. 🫶

👑 AUTHOR 👑

The Cyrus Team

💖 COPYRIGHT 💖

1993–2026, The Cyrus Team

and that's a wrap on IMAPD.CONF(5) — the whole thing, every last option, no crumbs left behind 💅✨

This one's from the 3.12.4 era, dated Sep 09, 2026 — so you know it's fresh. Go forth and configure, you absolute icon. slay 🦋🎀💜

📦 where this page comes from

fedora 44: cyrus-imapd 3.12.4-1.fc44 · BSD-Attribution-HPND-disclaimer
💖 girlified by deepseek-ai/DeepSeek-V4.1-Flash on 2026-10-18: a language model's rewrite of the original, under its terms

IMAPD.CONF(5)

💖 IMAPD.CONF(5) 💖

✨ NAME ✨

🌸 DESCRIPTION 🌸

🦋 FIELD DESCRIPTIONS 🦋

🎀 annotation_callout: <none>

🎀 annotation_callout_disable_append: 0

🎀 annotation_enable_legacy_commands: 0

🎀 aps_topic: <none>

🎀 aps_topic_caldav: <none>

🎀 aps_topic_carddav: <none>

🎀 aps_expiry: 1d

🎀 archive_enabled: 0

🎀 archive_days: <none>

🎀 archive_after: 7d

🎀 archive_maxsize: 1024 K

🎀 archive_keepflagged: 0

🎀 archivepartition-name: <none>

🎀 auditlog: 0

🎀 auth_mech: unix

🎀 autocreateinboxfolders: <none>

🎀 autocreatequota: <none>

🎀 autocreatequotamsg: -1

🎀 autosievefolders: <none>

🎀 generate_compiled_sieve_script: 0

🎀 autocreate_sieve_compiled_script: <none>

🎀 autosubscribeinboxfolders: <none>

🎀 autosubscribesharedfolders: <none>

🎀 autosubscribe_all_sharedfolders: 0

🎀 autocreate_acl: <none>

🎀 autocreate_inbox_folders: <none>

🎀 autocreate_post: 0

🎀 autocreate_quota: -1

🎀 autocreate_quota_messages: -1

🎀 autocreate_sieve_folders: <none>

🎀 autocreate_sieve_script: <none>

🎀 autocreate_sieve_script_compile: 0

🎀 autocreate_sieve_script_compiled: <none>

🎀 autocreate_subscribe_folders: <none>

🎀 autocreate_subscribe_sharedfolders: <none>

🎀 autocreate_subscribe_sharedfolders_all: 0

🎀 autocreate_users: anyone

🎀 autoexpunge: 0

🎀 backuppartition-name: <none>

🎀 backup_compact_minsize: 0

🎀 backup_compact_maxsize: 0

🎀 backup_compact_work_threshold: 1

🎀 backup_staging_path: <none>

🎀 backup_retention_days: <none>

🎀 backup_retention: 7d

backup_db: twoskip

backup_db_path: <none>

backup_keep_previous: 0

boundary_limit: 1000

caldav_accept_invalid_rrules: 0

caldav_alarm_db_path: <none>

caldav_alarm_support_components: VEVENT VTODO

caldav_alarm_suppress_file: <none>

caldav_allowattach: 1

caldav_allowcalendaradmin: 0

caldav_allowscheduling: on

caldav_create_attach: 1

caldav_create_default: 1

caldav_create_sched: 1

caldav_historical_age: 7d

caldav_maxdatetime: 20380119T031407Z

caldav_mindatetime: 19011213T204552Z

caldav_realm: <none>

calendarprefix: #calendars

calendar_default_displayname: personal

calendar_user_address_set: <none>

calendar_component_set: VEVENT VTODO VJOURNAL VFREEBUSY VAVAILABILITY VPOLL

calendar_minimum_alarm_interval: 5m

calendar_suppress_duplicate_alarms: 1

carddav_allowaddmember: 0

carddav_allowaddressbookadmin: 0

carddav_realm: <none>

carddav_repair_vcard: 0

chatty: 0

client_bind: 0

🎀 `annotation_callout`: `<none>`

🎀 `annotation_callout_disable_append`: `0`

🎀 `annotation_enable_legacy_commands`: `0`

🎀 `aps_topic`: `<none>`

🎀 `aps_topic_caldav`: `<none>`

🎀 `aps_topic_carddav`: `<none>`

🎀 `aps_expiry`: `1d`

🎀 `archive_enabled`: `0`

🎀 `archive_days`: `<none>`

🎀 `archive_after`: `7d`

🎀 `archive_maxsize`: `1024 K`

🎀 `archive_keepflagged`: `0`

🎀 `archivepartition-`name: `<none>`

🎀 `auditlog`: `0`

🎀 `auth_mech`: `unix`

🎀 `autocreateinboxfolders`: `<none>`

🎀 `autocreatequota`: `<none>`

🎀 `autocreatequotamsg`: `-1`

🎀 `autosievefolders`: `<none>`

🎀 `generate_compiled_sieve_script`: `0`

🎀 `autocreate_sieve_compiled_script`: `<none>`

🎀 `autosubscribeinboxfolders`: `<none>`

🎀 `autosubscribesharedfolders`: `<none>`

🎀 `autosubscribe_all_sharedfolders`: `0`

🎀 `autocreate_acl`: `<none>`

🎀 `autocreate_inbox_folders`: `<none>`

🎀 `autocreate_post`: `0`

🎀 `autocreate_quota`: `-1`

🎀 `autocreate_quota_messages`: `-1`

🎀 `autocreate_sieve_folders`: `<none>`

🎀 `autocreate_sieve_script`: `<none>`

🎀 `autocreate_sieve_script_compile`: `0`

🎀 `autocreate_sieve_script_compiled`: `<none>`

🎀 `autocreate_subscribe_folders`: `<none>`

🎀 `autocreate_subscribe_sharedfolders`: `<none>`

🎀 `autocreate_subscribe_sharedfolders_all`: `0`

🎀 `autocreate_users`: `anyone`

🎀 `autoexpunge`: `0`

🎀 `backuppartition-`name: `<none>`

🎀 `backup_compact_minsize`: `0`

🎀 `backup_compact_maxsize`: `0`

🎀 `backup_compact_work_threshold`: `1`

🎀 `backup_staging_path`: `<none>`

🎀 `backup_retention_days`: `<none>`

🎀 `backup_retention`: `7d`

`backup_db: twoskip`

`backup_db_path: <none>`

`backup_keep_previous: 0`

`boundary_limit: 1000`

`caldav_accept_invalid_rrules: 0`

`caldav_alarm_db_path: <none>`

`caldav_alarm_support_components: VEVENT VTODO`

`caldav_alarm_suppress_file: <none>`

`caldav_allowattach: 1`

`caldav_allowcalendaradmin: 0`

`caldav_allowscheduling: on`

`caldav_create_attach: 1`

`caldav_create_default: 1`

`caldav_create_sched: 1`

`caldav_historical_age: 7d`

`caldav_maxdatetime: 20380119T031407Z`

`caldav_mindatetime: 19011213T204552Z`

`caldav_realm: <none>`

`calendarprefix: #calendars`

`calendar_default_displayname: personal`

`calendar_user_address_set: <none>`

`calendar_component_set: VEVENT VTODO VJOURNAL VFREEBUSY VAVAILABILITY VPOLL`

`calendar_minimum_alarm_interval: 5m`

`calendar_suppress_duplicate_alarms: 1`

`carddav_allowaddmember: 0`

`carddav_allowaddressbookadmin: 0`

`carddav_realm: <none>`

`carddav_repair_vcard: 0`

`chatty: 0`

`client_bind: 0`

`client_bind_name: <none>`

`client_timeout: 10s`

`commandmintimer: <none>`

`configdirectory: <none>`

`createonpost: 0`