db1108, the analytics misc host needs to be upgraded to Bullseye.
This can be done anytime as the mariadb version isn't changing (10.4)
Reminder: shutdown mariadb before going for the reimage
Description
Status | Subtype | Assigned | Task | ||
---|---|---|---|---|---|
Unknown Object (Task) | |||||
Resolved | Papaul | T326661 Q3:rack/setup/install db1207-db1225 | |||
Resolved | Marostegui | T326669 Productionize db1206-db1225 | |||
Open | None | T291916 Tracking task for Bullseye migrations in production | |||
Resolved | Marostegui | T298585 Upgrade WMF database-and-backup-related hosts to bullseye | |||
Restricted Task | |||||
Restricted Task | |||||
Restricted Task | |||||
Restricted Task | |||||
Declined | BTullis | T304492 Upgrade db1108 to Bullseye |
Event Timeline
I'll carry out this upgrade, although it might be in a few weeks' time. Hope that's OK.
Hey @Marostegui,
Is this work blocking anything at the moment? It currently on our radar, but not prioritised. Please let me know if its a blocker or urgent and we can make sure it gets prioritised accordingly?
It is not blocking anything specifically, but this host needs to _at least_ be rebooted soon as there are several kernel upgrades it has missed as it has been up for 650 days cc @MoritzMuehlenhoff
I rebooted db1108, but the systemd service definitions for the analytics_meta and matomo database instances no longer exist for some reason. Checking the cause now.
Oops. This was an error on my part. I was looking in the wrong place for the systemd units. Thanks for correcting me @Marostegui
OK, so that's a [well overdue] reboot completed. As @EChetty said we've struggled to find the time to reimage this host, but will try to do so soon.
We have a lot of other buster->bullseye reimaging to do in the next few months, so hopefully we will be able to fit this in as well.
I would prefer if this is given a bit of a higher priority. This task has been here for a year and we are no longer providing updates to the buster Mariadb package. Only bullseye ones.
Understood. Thanks @Marostegui - Elevated the priority as requested. This is a three week sprint for our team so I'm hoping to do it very soon.
Thanks - I'm happy to help with the process. Basically it's just to double check the host is in the right partman recipe so /srv doesn't get wiped.
Then stop mysql, reimage, and start mysql :)
db1108 will need to be replaced by the new hardware (T326669), which will be installed on bullseye, so probably we can just migrate the data to that new host.