We're testing Debian Trixie in our DB infra, so far we've tested it on db*, es* and currently doing dbproxy*
I believe we should go ahead and get one of the new hosts at T409162 to be reimaged with Trixie before we get them with data, so we know if everything will work out of the box
Description
Description
| Status | Subtype | Assigned | Task | ||
|---|---|---|---|---|---|
| Resolved | Marostegui | T406981 Compile and test MariaDB 10.11 on Debian Trixie | |||
| Resolved | Request | Jhancock.wm | T408407 decommission es2028 | ||
| Resolved | Marostegui | T407472 Install a testing db with Debian Trixie | |||
| Unknown Object (Task) | |||||
| Open | None | T409162 Q2:rack/setup/install clouddb1026-1033 | |||
| Open | None | T415165 Install a clouddb hosts with Debian Trixie |
Event Timeline
Comment Actions
If this goes well, any reason not to install all of those on Trixie from the start to avoid having to re-image them later?
Comment Actions
I would suggest:
- reimage one host
- install mariadb and set up replication on that host
- if it all goes well, reimage following hosts
fnegri moved this task from Inbox to Watching on the cloud-services-team board.
According to https://wikitech.wikimedia.org/wiki/Portal:Data_Services/Admin/Wiki_Replicas#Who_admins_what the reimage itself is done by WMCS, so I moved it to the wrong column.