[Maria-discuss] Page Compression in production and backup strategies
![](https://secure.gravatar.com/avatar/3d68a25b1c702749b0c7f84369433198.jpg?s=120&d=mm&r=g)
Hi folks, Does anyone here use page compression ( https://mariadb.com/kb/en/mariadb/compression/ ) in production? (Or know folks who do?) What prompted my question was seeing that XtraBackup is not supported when using page compression. Given the popularity of the XtraBackup, I'm left wondering if there's some other backup strategy people use (especially since it's also used to sync and provision slaves), or if page compression is simply not used by folks who have master/slave setups, or perhaps don't require backups? Searching has not been helpful in my finding the answer, other than a pull request ( https://github.com/percona/percona-xtrabackup/pull/200 ) by Jan Lindstrom from last year. Hoping you can help, best regards, Tim
![](https://secure.gravatar.com/avatar/359652f5d5394fdc76fea714c0e40c97.jpg?s=120&d=mm&r=g)
Am 10.04.2017 um 19:44 schrieb Tim Kersten:
Hi folks,
Does anyone here use page compression ( https://mariadb.com/kb/en/mariadb/compression/ ) in production? (Or know folks who do?)
What prompted my question was seeing that XtraBackup is not supported when using page compression. Given the popularity of the XtraBackup, I'm left wondering if there's some other backup strategy people use (especially since it's also used to sync and provision slaves), or if page compression is simply not used by folks who have master/slave setups, or perhaps don't require backups?
well, for init salves i just rsync the master hot, the stop the master and do a second rsync for at least a decade, the downtime is short and it's completly scriptable (and delete existing binlogs in case you re-init a slave, the whole process scritable too) and for backups i just shutdown a slave instance, rsync the datadir to the backup location and start the slave again - no tools need, just some lines of shellscripts
![](https://secure.gravatar.com/avatar/9444e6c4bbf35ad00c7ad475c9e9492d.jpg?s=120&d=mm&r=g)
Hi Tim, We use a slave to take backups from. So, stop the slave, take the backup using your tool of choice, restart the slave. Or, if you need the slave for some read only type uses perhaps, we use mylvmbackup to take a hot backup on the slave without needing to shut down mariadb slave. mylvmbackup is a free script, works well, and, requires lvm and uses snapshots. So, it’s a decent free choice for backups. The backup is easily restored, it’s just a tar archive. We use lz4 compression for great speed in the compression and minimal impact on the slave. It’s easy to restore a slave backup and even make it the master. Steve
On Apr 10, 2017, at 12:44 PM, Tim Kersten
wrote: Hi folks,
Does anyone here use page compression ( https://mariadb.com/kb/en/mariadb/compression/ ) in production? (Or know folks who do?)
What prompted my question was seeing that XtraBackup is not supported when using page compression. Given the popularity of the XtraBackup, I'm left wondering if there's some other backup strategy people use (especially since it's also used to sync and provision slaves), or if page compression is simply not used by folks who have master/slave setups, or perhaps don't require backups?
Searching has not been helpful in my finding the answer, other than a pull request ( https://github.com/percona/percona-xtrabackup/pull/200 ) by Jan Lindstrom from last year.
Hoping you can help,
best regards,
Tim
_______________________________________________ Mailing list: https://launchpad.net/~maria-discuss Post to : maria-discuss@lists.launchpad.net Unsubscribe : https://launchpad.net/~maria-discuss More help : https://help.launchpad.net/ListHelp
participants (3)
-
Reindl Harald
-
Sales
-
Tim Kersten