We have seen a number of users start upgrades without backups. The upgrade has broken, and they are not able to restore.
By default we should take a backup of the server before we proceed, to encourage admins to perform good behaviour, and so that we have a supportable fallback if something goes wrong.
triage notes:
ab: upgrade happens as part of RPM package update, we cannot expect ipa-backup to be usable at this stage (offline, etc) dkupka: Is ipa-backup/restore really good way to go? Isn't VM/FS-snapshot better? mkosek: it is not, but it is what it is pv: can we even do it? A backup might be quite large and creating a new set of issues. stn: what would happen when upgrade happens in replicated environment and some parts are already replicated? Do we say the rest of the topology needs to be disconnected? dkupka: replica is disconnected before upgrade so this should not be a problem stn: alright, thx mbasti: -1 H: -1, would introduce more issues than solve IMHO mbabinsk: -1 stn: -1 TODO: check that our current upgrade instructions recommend backup close as won't fix
Metadata Update from @firstyear: - Issue assigned to someone - Issue set to the milestone: 0.0 NEEDS_TRIAGE