pg and mysql paas databases should be backed up if they are not already.
As per bug 949005, "backup6.db.scl3.mozilla.com has been kickstarted, BIOS updated, puppetized, and basic nagios monitoring added. I did format the storage blade in RAID 6 just to make sure everything is operating, but is not currently mounted on the server as you might have other plans." This would be for the postgres backups.
Can you please confirm that the mysql databases for apps on the production paas are being backed up? What is the backup schedule?
I can confirm MySQL is not currently being backed up, but I can fast-track it for this week. The main limitation is the netflow, which I'll request now.
r 82268 and 82269 have puppet updates for this backup, will initially live on backup4 in scl3.
mysql paas is being backed up. As with everything else in this data center, backups are kept for 3 days.
The backups schedule is once daily there is a logical backup (sql export, one per database) and once daily there is a raw backup (file copy). This way we can restore one database as easily as we can restore all of them, and we have 2 copies of the database in different formats in case one does not work for some reason. We also have an HA setup for both MySQL and Postgres in the PaaS - we have a replicated copy of the master to the slave. So if the MySQL master dies, we have options. Does this address your needs? If you have others, please let me know.
Thanks for the quick response, Sheeri. That sounds great to me. :)
If you could write up documentation on this, I can test it out by backing up the postgres paas databases.
Added user access from backup6.db.scl3 to puppet in revision 86764
Added node backup6.db.scl3 in revision 86755 to dbbackups.pp
Weekly pg_dumps and daily base backups running successfully on backup6.db.scl3