ctl_backups - Man Page

ctl_backups — Cyrus IMAP documentation

Perform administrative operations directly on Cyrus backups.

ctl_backups [OPTIONS] compact [MODE] backup...
ctl_backups [OPTIONS] list [LIST OPTIONS] [[MODE] backup...]
ctl_backups [OPTIONS] lock [LOCK OPTIONS] [MODE] backup
ctl_backups [OPTIONS] reindex [MODE] backup...
ctl_backups [OPTIONS] stat [MODE] backup...
ctl_backups [OPTIONS] verify [MODE] backup...

NOTE:

Cyrus Backups are experimental, incomplete, and deprecated as of 3.10.

ctl_backups is a tool for performing administrative operations on Cyrus backups.

ctl_backups reads its configuration options out of the imapd.conf(5) file unless specified otherwise by -C.

In all invocations, backup is interpreted according to the specified MODE. See Modes below.

ctl_backups provides the following sub-commands:

compact

Reduce storage required by the named backups.  Compact behaviour is influenced by the backup_compact_minsize, backup_compact_maxsize, backup_compact_work_threshold, and backup_retention_days configuration settings.  See imapd.conf(5) for details.

This should generally be invoked regularly, such as by adding an entry to the EVENTS section of cyrus.conf(5).  See Examples for an example.

If the backup_keep_previous configuration setting is enabled, compact will preserve the original data and index files (renaming them with a timestamp).  This is useful for debugging.

list

List backups.  See List Options for options specific to the list sub-command.  Columns are separated by tabs, and are:

• end time of latest chunk
• size of backup data file on disk
• userid to which the backup belongs
• path to backup data file

If no mode or backups are specified, lists all known backups (as if invoked with the -A mode).

lock

Obtain and hold a lock on the named backup.  Useful for operating on Cyrus backup files using non-Cyrus tools (such as UNIX tools or custom scripts) in relative safety.  See Lock Options for details.

reindex

Rebuild the indexes for the named backups, based on the raw backup data. This is useful if their index files have been corrupted, or if the index format has changed.

If the backup_keep_previous configuration setting is enabled, reindex will preserve the original index file (renaming it with a timestamp).  This is useful for debugging.

stat

Display stats for the named backups.  Columns are separated by tabs, and are:

• userid or filename
• compressed (i.e. on disk) size
• uncompressed size
• compactable size
• compression ratio
• utilisation ratio
• start time of latest chunk
• end time of latest chunk

The compactable size is an approximation of how much uncompressed data would remain after compact is performed.  The utilisation ratio is this figure expressed as a percentage of the uncompressed size.  Note that this approximation is an underestimate.  That is to say, a backup that has just been compacted will probably still report less than 100% utilisation.

verify

Verify consistency of the named backups by performing deep checks on both the raw backup data and its index.

-C config-file

Use the specified configuration file config-file rather than the default imapd.conf(5).

-F,  --force

Force the operation to occur, even if it is determined to be unnecessary. This is mostly useful with the compact sub-command.

-S,  --stop-on-error

Stop-on-error.  With this option, if a sub-command fails for any particular backup, ctl_backups will immediately exit with an error, without processing further backups.

The default is to log the error, and continue with the next backup.

-V,  --no-verify

Don't verify backup checksums for read-only operations.

The read-only operations list and stat will normally perform a "quick" verification of the backup file being read, which checks the checksums of the most recent chunk.  This can be slow for backups whose most recent backup chunk is very large.

With this option, the verification step will be skipped.

-j,  --json

Produce output in JSON format.  The default is plain text.

-v,  --verbose

Increase the verbosity.  Can be specified multiple times.

-w,  --wait-for-locks

Wait for locks.  With this option, if a backup named on the command line is locked, execution will block until the lock becomes available.

The default is to skip backups that are currently locked.

Options that apply only to the list sub-command.

-t [hours], --stale[=hours]

List stale backups only, that is, backups that have received no updates in hours.  If hours is unspecified, it defaults to 24.

Options that apply only to the lock sub-command.

-c,  --create

Exclusively create the named backup while obtaining the lock.  Exits immediately with an error if the named backup already exists.

When the lock is successfully obtained, continue as per the other options.

-p,  --pause

Locks the named backup, and then waits for EOF on the standard input stream.  Unlocks the backup and exits once EOF is received.  This is the default mode of operation.

-s,  --sqlite3

Locks the named backup, and with the lock held, opens its index file in the sqlite3(1) program.  The lock is automatically released when sqlite3 exits.

-x command, --execute=command

Locks the named backup, and with the lock held, executes command using /bin/sh (as per system(3)).  The lock is automatically released when command completes.

The filenames of the backup data and index are made available to command in the environment variables $ctl_backups_lock_data_fname and $ctl_backups_lock_index_fname, respectively.

-A,  --all

Run sub-command over all known backups.

Known backups are recorded in the database specified by the backup_db and backup_db_path configuration options.

-D,  --domains

Backups specified on the command line are interpreted as domains.  Run sub-command over known backups for users in these domains.

-P,  --prefixes

Backups specified on the command line are interpreted as userid prefixes. Run sub-command over known backups for users matching these prefixes.

-f,  --filenames

Backups specified on the command line are interpreted as filenames.  Run sub-command over the matching backup files.  The backup files do not need to be known about in the backups database.

-m,  --mailboxes

Backups specified on the command line are interpreted as mailbox names. Run sub-command over known backups containing these mailboxes.

-u,  --userids

Backups specified on the command line are interpreted as userids.  Run sub-command over known backups for matching users.

This is the default if no mode is specified.

Scheduling ctl_backups compact to run each morning using the EVENTS section of cyrus.conf(5):

EVENTS {
    checkpoint    cmd="ctl_cyrusdb -c" period=30

    compact       cmd="ctl_backups compact -A" at=0400
}

imapd.conf(5), sqlite3(1), system(3)

The Cyrus Team

1993–2024, The Cyrus Team