Last modified: 2014-06-27 14:16:12 UTC
Make database, redis, and file system back a bit more robust (keep more than just last instance) Subtasks * communicate to labs admins that we're doing this * write script to put daily backups wherever labs says to put them * test full recover from backup on staging, no downtime Notes: * We need to backup: * database * redis * generated files * Each backup is expected to be ~3GB * Total per month ~90GB * We're not guarding against hardware failure * need to have at least 1 month of data, better a quarter as no data -> no dashboard, we would need to run all reports again
Change 139557 had a related patch set uploaded by Milimetric: Add backup role and scripts https://gerrit.wikimedia.org/r/139557
Change 139558 had a related patch set uploaded by Milimetric: Enable the new debug role if set https://gerrit.wikimedia.org/r/139558
tested by nuria in staging
Change 139557 merged by Ottomata: Add backup role and scripts to wikimetrics https://gerrit.wikimedia.org/r/139557
Change 139558 merged by Ottomata: Enable the new backup role in wikimetrics if set https://gerrit.wikimedia.org/r/139558