Monday, 2026-02-09

*** mhen_ is now known as mhen02:26
noonedeadpunk#startmeeting freezer15:03
opendevmeetMeeting started Mon Feb  9 15:03:47 2026 UTC and is due to finish in 60 minutes.  The chair is noonedeadpunk. Information about MeetBot at http://wiki.debian.org/MeetBot.15:03
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.15:03
opendevmeetThe meeting name has been set to 'freezer'15:03
noonedeadpunk#topic rollcall15:03
noonedeadpunko/15:03
mheno/15:03
noonedeadpunksorry for being late15:03
noonedeadpunk#topic elasticsearch engine15:06
noonedeadpunkSo, patches were pushed, as I did not get any negative feedback around them, I guess I will go along with merging them later today15:06
noonedeadpunkpretty much the last chance to vote against it :p15:07
mhenI mean many other OpenStack services do also use SQLAlchemy and they tend to have much higher processing load than I could imagine for Freezer so I think it should be fine15:09
noonedeadpunkyup, agreed15:09
noonedeadpunk#topic office hours15:09
noonedeadpunkI actually don't have much for agenda today15:09
noonedeadpunkas I think most patches and work has been merged15:09
noonedeadpunkI did not have much progress with central scheduling last week15:10
noonedeadpunkAnd will likely have negative time for it this week15:10
noonedeadpunkthough feature freeze is Feb 23...15:10
noonedeadpunkso this feels kinda super tight...15:11
Luzio/15:11
noonedeadpunkdoesn't mean I'm gonna stop or anything, but realistically it could mean that we won't land it for 2026.1... unless will backport it...15:11
noonedeadpunkwhich is not a suitable thing to backport to be frank15:12
mhenyea sounds like a pretty major architectural change15:12
mhenLuzi and I are currently involved in some other project so sadly we cannot offer any substantial contribution power the next few weeks either :/15:13
Luziyeah :/15:13
noonedeadpunksure, no worries at al15:14
noonedeadpunkhow are you plans to add some kind of resource snapshots?15:14
noonedeadpunkare they still valid or postponed/abandoned?15:14
mhenwhile working on all the other patchsets recently I had an increasingly hard time imagining how I could fit it into the architecture of Freezer to be completely honest15:15
mhenespecially tying all the individual backups (modes/engines) together and recreate the hierarchy of the metadata snapshot of a project15:16
noonedeadpunkI belive the mode/engines should be severely refactored to be frank, as I agree that's a mess today15:17
noonedeadpunkbut yeah15:17
noonedeadpunkI see where the challange is15:17
mhenideally, I would prefer a top-down approach: snapshot the metadata (topology) of a project (or filtered subsets of it) and then go from there to capture individual backup sources (server images, volumes etc.)15:18
noonedeadpunkfwiw, I've reached couple of users who tried to run freezer last year and reported issues with $things15:18
noonedeadpunkand got a positive feedback from one of them, that current master fixed most of their issues15:18
mhenoh cool15:18
noonedeadpunkbut then it can be done on API side?15:19
noonedeadpunknot sure though15:19
noonedeadpunkThe somehow interesting part is actually restoration part, as for network topology, user does not have enough permissions for full restoration15:22
noonedeadpunkspecifically part with external networks15:22
mhentracking the correct spots where individual backups belong in the whole topology is hard and backup metadata structure and contents differing between modes/engines makes it harder15:22
mhenyea, those parts would be placeholders that would need to be filled on restore15:23
noonedeadpunkfrom other side, freezer kinda does one architectural thing interestingly, as it runs with user privileges15:23
noonedeadpunkso it's abale to create vms/disks for the project without issues15:23
noonedeadpunkeven with central thing - it will be issuing keystone trust for auth15:24
noonedeadpunkso that should not be a problem at least15:24
mhenyes, that certainly helps for the project side of things15:24
mhenand I find the Keystone trust approach very good15:25
mhenbut I think at least that whole engine/mode thing needs a refactor before I would consider to pick up that project topology/metadata snapshot idea again15:27
noonedeadpunkok, let's see then how it may work out... Once scheduling part is ready I'll try to check on how to refactor these15:27
noonedeadpunkAnd for the next cycle there's hopefully some help coming from my org as well for this15:28
noonedeadpunkok, if that's it, I'd propose to end the meeting early15:29
mhen+115:29
noonedeadpunkfwiw, I'm not sure if I will be around next week.15:29
noonedeadpunkbut will let know 15:29
mhenack15:29
noonedeadpunk#endmeeting15:29
opendevmeetMeeting ended Mon Feb  9 15:29:38 2026 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)15:29
opendevmeetMinutes:        https://meetings.opendev.org/meetings/freezer/2026/freezer.2026-02-09-15.03.html15:29
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/freezer/2026/freezer.2026-02-09-15.03.txt15:29
opendevmeetLog:            https://meetings.opendev.org/meetings/freezer/2026/freezer.2026-02-09-15.03.log.html15:29

Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!