squatter - Cyrus IMAP documentation
Create SQUAT and Xapian indexes for mailboxes
general: squatter [ -C config-file ] [mode] [options] [source] i.e.: squatter [ -C config-file ] [ -v ] [ -a ] [ -S seconds ] [ -Z ] squatter [ -C config-file ] [ -v ] [ -a ] [ -i ] [ -N name ] [ -S seconds ] [ -r ] [ -Z ] mailbox... squatter [ -C config-file ] [ -v ] [ -a ] [ -i ] [ -N name ] [ -S seconds ] [ -r ] [ -Z ] -u user... squatter [ -C config-file ] [ -v ] [ -a ] -R [ -n channel ] [ -d ] [ -S seconds ] [ -Z ] squatter [ -C config-file ] [ -v ] [ -a ] -f synclogfile [ -S seconds ] [ -Z ] squatter [ -C config-file ] [ -v ] -t srctier(s)... -z desttier [ -F ] [ -U ] [ -T dir ] [ -X ] [ -o ] [ -S seconds ] [ -u user... ]
squatter creates a new text index for one or more IMAP mailboxes. The index is a unified index of all of the header and body text of each message in a given mailbox. This index is used to significantly reduce IMAP SEARCH times on a mailbox.
mode is one of indexer, search, rolling, synclog, compact or audit.
By default, squatter creates an index of ALL messages in the mailbox, not just those since the last time that it was run. The -i option is used to select incremental updates. Any messages appended to the mailbox after squatter is run, will NOT be included in the index. To include new messages in the index, squatter must be run again, or on a regular basis via crontab, an entry in the EVENTS section of cyrus.conf(5) or use rolling mode (-R).
In the first synopsis, squatter indexes all mailboxes.
In the second synopsis, squatter indexes the specified mailbox(es). The mailboxes are space-separated.
In the third synopsis, squatter indexes the specified user(s) mailbox(es).
For the latter two index modes (mailbox, user) one may optionally specify -r to recurse from the specified start, or -a to limit action only to mailboxes which have the shared /vendor/cmu/cyrus-imapd/squat annotation set to "true".
In the fourth synopsis, squatter runs in rolling mode. In this mode squatter backgrounds itself and runs as a daemon (unless -d is set), listening to a sync log channel chosen using the -n option, and set up using the sync_log_channels setting in imapd.conf(5). Very soon after messages are delivered or uploaded to mailboxes squatter will incrementally index the affected mailbox (see notes, below).
In the fifth synopsis, squatter reads a single sync log file and performs incremental indexing on the mailbox(es) listed therein. This is sometimes useful for cleaning up after problems with rolling mode.
In the sixth synopsis, squatter will compact indices from srctier(s) to desttier, optionally reindexing (-X) or filtering expunged records (-F) in the process. The optional -T flag may be used to specify a directory to use for temporary files. These files are eventually copied with rsync -a and then removed by rm. rsync can increase the load average of the system, especially when the temporary directory is on tmpfs. To throttle rsync it is possible to modify the call in imap/search_xapian.c and pass -\-bwlimit=<number> as further parameter. The -o flag may be used to direct that a single index be copied, rather than compacted, from srctier to desttier. The -u flag may be used to restrict operation to the specified user(s).
For all modes, the -S option may be specified, causing squatter to pause seconds seconds after each mailbox, to smooth loads.
When using the Xapian engine the -Z option may be specified, for the indexing modes. This tells squatter to consult the Xapian internally indexed GUIDs, rather than relying on what's stored in cyrus.indexed.db, allowing for recovery from broken cyrus.indexed.db at the sacrifice of efficiency.
squatter reads its configuration options out of the imapd.conf(5) file unless specified otherwise by -C.
- Only create indexes for mailboxes which have the shared
/vendor/cmu/cyrus-imapd/squat annotation set to "true".
The value of the /vendor/cmu/cyrus-imapd/squat annotation is inherited by all children of the given mailbox, so an entire mailbox tree can be indexed (or not indexed) by setting a single annotation on the root of that tree with a value of "true" (or "false"). If a mailbox does not have a /vendor/cmu/cyrus-imapd/squat annotation set on it (or does not inherit one), then the mailbox is not indexed. In other words, the implicit value of /vendor/cmu/cyrus-imapd/squat is "false".
- Missing documentation, starts audit mode. This feature is only available on the master branch.
- In rolling mode, don't background and do emit log messages on standard error. Useful for debugging. This feature was introduced in version 3.0.
- In compact mode, filter the resulting database to only include messages which are not expunged in mailboxes with existing name/uidvalidity. This feature was introduced in version 3.0.
- -f synclogfile
- Read the synclogfile and incrementally index all the mailboxes listed therein, then exit. This feature was introduced in version 3.0.
- Display this usage information.
- Incremental updates where indexes already exist.
- -N name
- Only index mailboxes beginning with name while iterating through the mailbox list derived from other options.
- In compact mode, if only one source database is selected, just copy it to the destination rather than compacting. This feature was introduced in version 3.0.
- Missing documentation, reindex parts mode. This feature is only available on the master branch.
- Run in rolling mode; squatter runs as a daemon listening to a sync log channel and continuously incrementally indexing mailboxes. See also -d and -n. This feature was introduced in version 3.0.
- Recursively create indexes for all sub-mailboxes of the user, mailboxes or mailbox prefixes given as arguments.
- -S seconds
- After processing each mailbox, sleep for "seconds" before continuing. Can be used to provide some load balancing. Accepts fractional amounts. This feature was introduced in version 3.0.
- -T directory
- When indexing, work on a temporary copy of the search engine databases in directory. That directory would typically be on some very fast filesystem, like an SSD or tmpfs. This option may not work with all search engines, but it's only effect is to speed up initial indexing. Xapian only. This feature was introduced in version 3.0.
- -t srctier...
- In compact mode, the comma-separated source tier(s) for the compacted indices. At least one source tier must be specified in compact mode. Xapian only. This feature was introduced in version 3.0.
- Extra options refer to usernames (e.g. email@example.com) rather than mailbox names. Usernames are space-separated. This feature was introduced in version 3.0.
- In compact mode, only compact if re-indexing. Xapian only. This feature is only available on the master branch.
- Increase the verbosity of progress/status messages. Sometimes additional messages are emitted on the terminal with this option and the messages are unconditionally sent to syslog. Sometimes messages are sent to syslog, only if -v is provided. In rolling and synclog modes, -vv sends even more messages to syslog.
- Reindex all the messages before compacting. This mode reads all the lists of messages indexed by the listed tiers, and re-indexes them into a temporary database before compacting that into place. Xapian only. This feature was introduced in version 3.0.
- -z desttier
- In compact mode, the destination tier for the compacted indices. This must be specified in compact mode. Xapian only. This feature was introduced in version 3.0.
- When indexing messages, use the Xapian internal cyrusid rather than referencing the ranges of already indexed messages to know if a particular message is indexed. Useful if the ranges get out of sync with the actual messages (e.g. if files on a tier are lost) Xapian only. This feature is only available on the master branch.
squatter is typically deployed via entries in cyrus.conf(5), in either the DAEMON or EVENTS sections.
For the older SQUAT search engine, which offers poor performance in rolling mode (-R) we recommend triggering periodic runs via entries in the EVENTS section, as follows:
Sample entries from the EVENTS section of cyrus.conf(5) for periodic squatter runs:
# reindex changed mailboxes (fulltext) approximately every three hours
squatter1 cmd="/usr/bin/ionice -c idle /usr/lib/cyrus/bin/squatter -i" period=180
# reindex all mailboxes (fulltext) daily
squattera cmd="/usr/lib/cyrus/bin/squatter" at=0117 }
For the newer Xapian search engine, and with sufficiently fast storage, the rolling mode (-R) offers advantages. Use of rolling mode requires that squatter be invoked in the DAEMON section.
Sample entries for the DAEMON section of cyrus.conf(5) for rolling squatter operation:
# run a rolling squatter using the default sync_log channel "squatter"
squatter cmd="squatter -R"
# run a rolling squatter using a specific sync_log channel
squatter cmd="squatter -R -n indexer" }
[NB: More examples needed]
Support for additional search engines was added in version 3.0.
The following command-line switches were added in version 3.0:
-F -R -X -d -f -o -u
The following command-line settings were added in version 3.0:
-S <seconds>, -T <directory>, -f <synclogfile>, -n <channel>, -t srctier..., -z desttier
The Cyrus Team, Nic Bernstein (Onlight)
1993-2018, The Cyrus Team
|March 9, 2021||3.2.6|