14:00:40 <abhishekk> #startmeeting glance
14:00:40 <opendevmeet> Meeting started Thu Aug  5 14:00:40 2021 UTC and is due to finish in 60 minutes.  The chair is abhishekk. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:00:40 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:00:40 <opendevmeet> The meeting name has been set to 'glance'
14:00:46 <abhishekk> #topic roll call
14:00:49 <redrobot> \o
14:00:52 <abhishekk> #link https://etherpad.openstack.org/p/glance-team-meeting-agenda
14:00:57 <abhishekk> o/
14:01:20 <abhishekk> lets wait couple of minutes for others
14:02:03 <rosmaita> o/
14:02:12 <croelandt> o/
14:02:15 <dansmith> o./
14:02:23 <abhishekk> cool, lets start
14:02:29 <jokke_> o/
14:02:34 <abhishekk> #topic release/periodic jobs update
14:02:41 <abhishekk> M3 4 weeks from now
14:02:45 <abhishekk> glance_store xena release - 2 weeks from now
14:03:05 <abhishekk> We don't have much from store side now
14:03:37 <abhishekk> Periodic jobs 2 time outs this week, trying to figure out the problem but don't have much time atm
14:03:56 <abhishekk> #topic M3 targets
14:04:10 <abhishekk> Policy refactoring work
14:04:17 <abhishekk> Finally things are rolling
14:04:37 <abhishekk> thanks to croelandt and lbragstad for helping us to move ahead
14:04:54 <abhishekk> You will get total overview of refactoring work from below sheet
14:05:04 <abhishekk> #link  https://docs.google.com/spreadsheets/d/1SWBq0CsHw8jofHxmOG8QeZEX6veDE4eU0QHItOu8uQs/edit?pli=1#gid=73773117
14:05:21 <abhishekk> Still long way to go..
14:05:48 <abhishekk> Any questions related to policy work?
14:05:48 <dansmith> yep, but a lot of momentum already I think
14:05:59 <abhishekk> ++
14:06:33 <abhishekk> I take that as know, moving ahead
14:06:37 <abhishekk> Cache API
14:06:46 <abhishekk> we have new revision up for glance side work
14:07:02 <abhishekk> #link https://review.opendev.org/c/openstack/glance/+/792022
14:07:20 <abhishekk> Need to have close look on this, will do it today after the meeting
14:07:31 <abhishekk> Pending is now client side work
14:08:02 <jokke_> I have a client patch up as well
14:08:22 <abhishekk> So all reviewers spend some time to review
14:08:41 <abhishekk> That is still PoC but you can refer it
14:08:55 <jokke_> #link https://review.opendev.org/c/openstack/python-glanceclient/+/800172
14:09:14 <abhishekk> ack
14:09:15 <jokke_> yes, please provide feedback and lets move those on
14:09:43 <abhishekk> Noted, will look asap
14:09:59 <abhishekk> Metadef project persona integration
14:10:09 <abhishekk> #link  https://review.opendev.org/c/openstack/glance/+/798700/
14:10:18 <abhishekk> I think these are also in good shape
14:10:45 <abhishekk> Moving to next topic
14:10:58 <abhishekk> #topic Wallaby backports discussion (croelandt)
14:10:59 <abhishekk> 
14:11:07 <abhishekk> croelandt, stage is yours
14:11:42 <abhishekk> Ok, let me explain this on his behalf
14:11:47 <croelandt> yeah
14:11:48 <croelandt> sorry
14:12:01 <abhishekk> no problem, do you want me to continue ?
14:12:02 <croelandt> so we have those two patches we want to backport
14:12:16 <croelandt> the first one applies almost cleanly, the second one applies cleanly
14:12:17 * abhishekk go ahead
14:12:30 <croelandt> but only the second one is suitable for backport, since the first one is a new feature
14:12:45 <croelandt> *but* if we only apply the second one (a bugfix), then it conflicts like crazy
14:12:56 <croelandt> and to fix the conflicts I'm gonna have to basically rewrite the first one
14:13:05 <croelandt> so I'd be tempted to backport both or none of them
14:13:17 <croelandt> erf, Rajat isn't here, he knows more about the patches content than I do :/
14:13:21 <jokke_> What are we talking about? Any references?
14:13:37 <abhishekk> #link https://review.opendev.org/c/openstack/glance_store/+/782200
14:13:43 <abhishekk> this is 1st one
14:13:53 <abhishekk> #link https://review.opendev.org/c/openstack/glance_store/+/796577
14:13:57 <abhishekk> this is actual bug fix
14:13:59 <croelandt> yep :)
14:14:24 <rosmaita> so why exactly is 782200 a new feature?
14:14:35 <abhishekk> I think if the bug is important then we can backport supporting patches as well
14:14:46 <dansmith> the first one changes our interaction with cinder quite a bit right?
14:15:11 <croelandt> rosmaita: it uses the new attachment support from Cinder
14:15:20 <croelandt> which is also a required backport in Cinder from what I understand
14:15:23 <rosmaita> "new" == since Pike or something
14:15:58 <abhishekk> dansmith, right
14:16:00 <dansmith> seems a bit risky to me, despite that
14:16:11 <rosmaita> we already merged that change to wallaby, and broke grenade
14:16:17 <dansmith> the microversion it requires is much newer than pike, if I recall
14:16:18 <rosmaita> (grenade is fixed now)
14:16:28 <rosmaita> astually, that was a different break, forget what i said
14:16:46 <jokke_> I do agree, it's really not something that aligns with our stable policy for backport
14:17:05 <croelandt> jokke_: I'm fine with not doing the backport :;)
14:17:13 <abhishekk> I think I lost the connection
14:17:14 <abhishekk> ?
14:17:18 <croelandt> but keep in mind that if we want ot backport https://review.opendev.org/c/openstack/glance_store/+/796577
14:17:27 <jokke_> abhishekk: nope, at least tx happens still ;)
14:17:30 <croelandt> we basically are going to reimplement half of #782200
14:18:07 <dansmith> well, that patch also seems like you could argue it's too chunky for a backport
14:18:24 <dansmith> if it's something that didn't work before, it's a feature despite having a bug number attached :)
14:18:32 <croelandt> hehe
14:18:38 <abhishekk> :D
14:18:50 <abhishekk> croelandt, I think its better not to backport it upstream
14:18:52 <croelandt> we'll tell the distro maintainers to do their own backports then
14:18:57 <dansmith> ++
14:19:00 <croelandt> which is fine since I'm the distro maintainer *cough*
14:19:06 <abhishekk> :P
14:19:24 <abhishekk> anything else you have to add here ?
14:19:28 <dansmith> for us downstream, it's fine to take that risk and do the testing to ensure it.. we know what cinder people have, what order they will apply the updates, what their general config can be
14:19:38 <dansmith> but upstream it seems too risky, IMHO
14:19:45 <jokke_> dansmith: I tend to agree, also if we just want to block writing qcow into cinder nfs, I don't see how that is related with the new attachment API.
14:19:55 <dansmith> jokke_: agree
14:19:57 <croelandt> yeah, I'm fine with not backporting upstream
14:20:06 <jokke_> Maybe if this is important enought to backport it needs independent fix rather than backport
14:20:18 <dansmith> or some mitigation thing
14:20:26 <croelandt> jokke_: yeah, we need to figure out whether we want it fixed in W
14:20:57 <abhishekk> ack, that we can discuss on ML we already have
14:21:05 <abhishekk> moving ahead
14:21:08 <abhishekk> #topic FIPS CI Jobs (alee/dmendiza)
14:21:13 <redrobot> \o
14:21:18 * redrobot is aka dmendiza
14:21:21 <jokke_> it's kind of one of these things like "It's not gonna work and we can tell the user so early or they can find it out down the line" :P
14:21:43 <abhishekk> redrobot, go ahead
14:22:02 <redrobot> Yeah, so Ade and I are working on an effort to test all of OpenStack on FIPS-enabled systems
14:22:31 <redrobot> to that end we've been adding CI jobs to every project to run existing test suites in nodes which have had FIPS turned on
14:22:38 <redrobot> #link https://review.opendev.org/c/openstack/glance/+/790536
14:22:42 <redrobot> ^^ is the one for glance
14:22:58 <abhishekk> Can we do it early in next cycle?
14:23:29 <redrobot> Up to you :)  Obviously we'd rather do it sooner rather than later.
14:23:31 <abhishekk> Frankly speaking we are just 4 weeks away from 3rd and final milestone and I am not eager to increase gate timeline
14:23:58 <abhishekk> We can start step by step rather adding bunch of jobs at a time
14:23:59 <rosmaita> maybe make it a periodic job
14:24:11 <jokke_> I think there was patch proposed by Ade already that was green, so it's not like we're gonna need tons of work in Glance side to make it work. I think Brian had some concerns about it 'though
14:24:39 <redrobot> It's the same patch I linked
14:24:39 <abhishekk> jokke_, in latest patch I think Ade has added 4 to 5 jobs
14:25:05 <redrobot> Yeah, I think the initial revisions were not running enough of the test suite to find failures.
14:25:11 <abhishekk> #link https://review.opendev.org/c/openstack/glance/+/790536
14:25:16 <rosmaita> my concern was that it added code so you could run mysql/postgres in fips mode, but since we weren't actually using either of those, was hard to see that it worked
14:25:33 <jokke_> But looking the job list we have in the gate, I'd tend to lean towards periodic job too rather than gating job. At least we keep eye on them weekly so it wouldn't be something that gets ignored for years there
14:25:58 <redrobot> yeah, we had to patch mysql/postgres because there's a test script in your repo that installs it and sets up tables and what not.
14:26:10 <rosmaita> yeah, it would unblock ade and redrobot from finishing, and hten we could make it a regular job later
14:26:24 <abhishekk> to start with periodic job sounds better idea
14:26:40 * redrobot needs to freshen up on periodic jobs
14:26:47 <redrobot> Yeah, that sounds like a good start for us
14:26:47 <jokke_> I'm kind of leaning towards Does it ever need to be regular gating job
14:26:50 <jokke_> ?
14:27:13 <dansmith> periodic seems fine to me for this
14:27:17 <redrobot> jokke_ we would prefer that, yes.  It would be better to prevent regressions on every patch once we get it working
14:27:34 <redrobot> to be clear, we would prefer a gating job rather than a periodic
14:27:50 <redrobot> but we're OK with starting with a periodic job until it is passing consistenly
14:27:57 <jokke_> redrobot: I do undertand the concern, the reality just is, our gate job list is hot mess already consuming incredible amounts of resources
14:28:33 <abhishekk> right and at this moment I don't want to increase that time line
14:28:33 <redrobot> understood.  I'll discuss with Ade, but we'll plan for periodic jobs for now for sure
14:28:35 <dansmith> could we make one of our regular jobs fipsified once we get it all fixed?
14:28:46 <dansmith> meaning, is there any harm to other stuff we need to do?
14:29:02 <jokke_> Obviously if it looks like we're breaking it in weekly basis it would make sense to be gating job, but if it keeps going green month after month in periodic jobs, do we really need that resource hog there
14:29:04 <dansmith> if not, once we get it ready, we can just convert one of our special jobs like the cinder multistore one or something to use fips
14:29:06 <jokke_> ?
14:29:20 <rosmaita> dansmith: that sounds like a good idea, kill 2 birds with one stone
14:29:39 <abhishekk> I think we can, but early in next cycle ?
14:29:48 <redrobot> dansmith yeah, interesting idea.  In theory once we iron out any FIPS issues we find it should be pretty much the same tests.
14:29:51 <rosmaita> abhishekk: ++
14:29:55 <dansmith> for sure, after it's stable and after we're through any deadlines
14:30:07 <dansmith> redrobot: cool, then let's aim for periodic now, and integrated later
14:30:09 <jokke_> dansmith: that would make much more sense
14:30:09 <abhishekk> works for me
14:30:23 <redrobot> sounds good, y'all, thanks
14:30:29 <abhishekk> cool
14:30:42 <abhishekk> redrobot, let me know if you need any help in periodic jobs
14:30:57 <redrobot> abhishekk will do, thanks!
14:31:15 <abhishekk> moving to Open discussion
14:31:18 <rosmaita> redrobot: you may want to make it also experimental so you can run it on demand
14:31:30 <rosmaita> at least while you are developing it
14:31:34 <abhishekk> ++
14:31:52 <abhishekk> #topic Open discussion
14:32:11 <jokke_> #link https://review.opendev.org/c/openstack/glance/+/802868
14:32:15 <abhishekk> Should we have additional weekly sync up for policy refactoring ???
14:32:33 <jokke_> #link https://review.opendev.org/c/openstack/glance/+/802867
14:32:57 <jokke_> low potential for merge conflicts and quick reviews, thanks! ^^
14:33:29 <abhishekk> I think we are getting close to 3rd milestone and we should sync to discuss blocker, progress once in a week >
14:33:55 <abhishekk> I will try to get lbragstad's time for the same
14:34:05 <jokke_> #link https://review.opendev.org/c/openstack/glance_store/+/800227 quick bugfix for glance_store too before we hit final release deadline
14:34:25 <dansmith> abhishekk: I guess it feels like we have good momentum right now, and not many blockers
14:34:40 <dansmith> so I don't feel like we _need_ the sync up, but if you do, then I'm happy to participate
14:34:53 <abhishekk> dansmith, may be we should revisit this next week
14:35:03 <abhishekk> ack, thank you
14:35:13 <dansmith> sounds good
14:35:23 <jokke_> #link https://review.opendev.org/c/openstack/glance/+/800101 rosmaita smcginnis stable could do with some love too
14:36:01 <abhishekk> cool, that's it from me today
14:36:32 <jokke_> #link https://review.opendev.org/c/openstack/glance/+/800100 abhishekk rosmaita smcginnis another backport
14:36:39 <abhishekk> jokke_, ack
14:36:40 <rosmaita> :)
14:37:30 <abhishekk> If nothing else then we can wrap up and utilize remaining time in reviews :D
14:37:33 <pdeore> #link https://review.opendev.org/c/openstack/glance/+/775968 pending for reviews since quite a long time :)
14:38:27 <abhishekk> ack
14:38:34 <jokke_> Also as the glance_store release deadline is the first one we're going to hit, would be great to make sure we get everything in soon what we want in there
14:38:34 <abhishekk> I take that as no
14:39:03 <jokke_> just my 2 cents in your local currency
14:39:16 <abhishekk> jokke_, ack, thank you
14:39:29 <abhishekk> Thank you all, have a nice weekend
14:39:34 <rosmaita> ty
14:39:39 <jokke_> TY
14:39:50 <abhishekk> Ty
14:39:53 <pdeore> Thanks!!
14:40:01 <abhishekk> #endmeeting