14:00:02 <abhishekk> #startmeeting glance
14:00:02 <opendevmeet> Meeting started Thu Nov 18 14:00:02 2021 UTC and is due to finish in 60 minutes.  The chair is abhishekk. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:00:02 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:00:02 <opendevmeet> The meeting name has been set to 'glance'
14:00:03 <abhishekk> #topic roll call
14:00:07 <abhishekk> #link https://etherpad.openstack.org/p/glance-team-meeting-agenda
14:00:09 <abhishekk> o/
14:00:46 <abhishekk> waiting for others to join
14:02:22 <pdeore> o/
14:02:38 <abhishekk> lets wait couple of minutes more
14:02:51 <rosmaita> o/
14:04:09 <abhishekk> Ok, lets start, others may/can join in between
14:04:22 <abhishekk> #topic Announcement
14:04:37 <abhishekk> OpenInfra summit is back
14:04:49 <rosmaita> do you know whether the PTG will be there, too?
14:04:54 <abhishekk> So in person OpenInfra events are back
14:05:03 <abhishekk> at the moment there is only summit
14:05:09 <rosmaita> ok
14:05:12 <abhishekk> I think PTG is still under discussion
14:05:32 <abhishekk> For early registration you can use,  https://openinfra.dev/summit/
14:05:51 <abhishekk> It is between 7 to 9 June and in Berlin
14:06:16 <abhishekk> Hopefully we will be able to meet in person this time
14:06:25 <jokke_> fingers crossed
14:06:30 <abhishekk> ++
14:06:35 <abhishekk> Moving ahead
14:06:41 <abhishekk> #topic release/periodic jobs update
14:06:52 <abhishekk> This is milestone 1 release week
14:07:02 <abhishekk> and we are not releasing anything this time
14:07:08 <abhishekk> So we are good \o/
14:07:08 <jokke_> :P
14:07:33 <jokke_> I guess our m1 targets didn't make it ;)
14:07:54 <abhishekk> :D
14:07:57 <abhishekk> AND periodic jobs all are green as well
14:08:03 <jokke_> uuh
14:08:13 <jokke_> that's cool
14:08:25 <abhishekk> I saw couple of timeouts on glance patches this week, but that is due to heavy queue in the gate
14:08:40 <abhishekk> Moving to next topic
14:08:48 <abhishekk> #topic Catching up with reviews and specs
14:09:00 <abhishekk> We have couple of specs up for reviews
14:09:14 <abhishekk> So requesting all of you to kindly provide your feedback on them
14:09:43 <abhishekk> AFAIK, Quota usage API, metadef tag append and resource type modification specs are up
14:10:16 <abhishekk> Please put some time in reviewing them
14:10:36 <abhishekk> moving ahead
14:10:40 <Mitya_Eremeev> sorry
14:10:56 <Mitya_Eremeev> Could you disscuss my patch ?
14:11:10 <abhishekk> Mitya_Eremeev, your topic is next so we can discuss it at that time
14:11:18 <Mitya_Eremeev> sorry)))
14:11:23 <abhishekk> no problem
14:11:36 <abhishekk> #topic Secure RBAC - Impact, what's next?
14:12:08 <abhishekk> lbragstad, if you are around, can you put some light on current Yoga goal for RBAC
14:13:08 <abhishekk> Does addition of project-manager role has any impact for glance project personas
14:13:52 <rosmaita> i thought project-manager was a stretch goal for Yoga?
14:14:34 <abhishekk> yeah, I think so as well
14:14:43 <rosmaita> here's the proposal: https://review.opendev.org/c/openstack/governance/+/815158
14:14:52 <rosmaita> i haven't read the latest version, but looks like dansmith has
14:15:11 <abhishekk> right, and it has two specific targets
14:15:24 <abhishekk> 1 is implementation of project-manager role
14:15:34 <abhishekk> 2 openstack client compatibility
14:15:37 <lbragstad> o/
14:15:45 <lbragstad> i'm catching up on that today
14:15:48 <abhishekk> cool, here he is
14:15:53 <lbragstad> and addressing feedback
14:16:28 <lbragstad> but the biggest change from previous versions is that,
14:16:34 <abhishekk> lbragstad, ack, I think we can sync some time early next week (if you have time)
14:16:45 <lbragstad> project-admin will remain an operator specific persona
14:18:15 <abhishekk> ok
14:18:38 <abhishekk> when is next RBAC open office hours meeting?
14:18:48 <lbragstad> the system personas are truly reserved for system-specific APIs (which I don't think glance has?)
14:18:59 * lbragstad checks
14:19:17 <abhishekk> right
14:19:23 <lbragstad> the TC is discussing the goal today
14:19:27 <lbragstad> in about 3 hours
14:19:43 <abhishekk> Ok, I will try to be there to clear some doubts
14:19:45 <abhishekk> thank you
14:19:50 <lbragstad> in 2.5 hours
14:19:52 <lbragstad> actually
14:20:09 <abhishekk> and also if you have some time, we can sync as per your availability as well
14:20:15 <lbragstad> ++ sounds good
14:20:19 <abhishekk> next week *
14:20:29 <abhishekk> thank you, please let me know
14:20:40 <lbragstad> i'll be around all day monday - wednesday
14:20:44 <lbragstad> so - i can make any time work
14:20:59 <abhishekk> Great, I will ping you on monday then
14:21:10 <abhishekk> moving ahead
14:21:27 <abhishekk> #topic Discussion of spec lite
14:21:46 <abhishekk> Add ability to purge all needed rows by glance-manage script
14:21:52 <abhishekk> Mitya_Eremeev, stage is yours
14:22:00 <abhishekk> just to put some light
14:22:04 <Mitya_Eremeev> thanks
14:22:23 <abhishekk> Mitya_Eremeev, reported a bug for the same, and as it is enhancement I have asked him for spec lite
14:22:35 <abhishekk> #link https://review.opendev.org/c/openstack/glance-specs/+/817938
14:22:51 <Mitya_Eremeev> https://review.opendev.org/c/openstack/glance/+/813691
14:23:11 <abhishekk> I do think we can have one option max-rows for glance db purge operation
14:23:17 <abhishekk> Mitya_Eremeev, have you tried that ?
14:23:59 <Mitya_Eremeev> max-rows have nothing in common with number of all deleted rows
14:24:39 <Mitya_Eremeev> it's just some limitation how many rows will be deleted by script
14:24:48 <abhishekk> ok
14:25:06 <abhishekk> so you just want to delete/purge all rows in one command
14:25:12 <Mitya_Eremeev> so no guarantess that all deleted rows will be purged
14:25:40 <Mitya_Eremeev> Yes, I think user should have ability to purge in one pass
14:26:11 <jokke_> So currently we're checking that max-rows is at least one. We could be consisten and use -1 there
14:26:25 <abhishekk> ack, so we have a spec lite up for the same, please review it and give some early feedback to Mitya_Eremeev
14:26:25 <Mitya_Eremeev> otherwise user wil do some monkey job just by launching script many times until script deleted all
14:26:45 <abhishekk> so -1 means delete all ?
14:26:52 <Mitya_Eremeev> no
14:27:09 <jokke_> abhishekk: that's how our other limits works
14:27:40 <abhishekk> Mitya_Eremeev, he is giving suggestion
14:27:42 <jokke_> so yeah, --max-rows -1 and --age-in-days 0 would equal to purge all
14:28:06 <Mitya_Eremeev> if pass key --purhe_all then --max row is just number of rows in one step
14:28:25 <jokke_> no need for an extra parameter for that nor breaking anyone
14:28:49 <Mitya_Eremeev> I don't think that max-rows allows negative values
14:29:24 <jokke_> Mitya_Eremeev: that's what I'm saying .... don't add the extra --purhe-all, but allow -1 as --max-rows
14:29:33 <Mitya_Eremeev> understood
14:29:35 <abhishekk> Mitya_Eremeev, we can modify it to allow
14:29:47 <Mitya_Eremeev> but we should not do that
14:29:52 <abhishekk> jokke_, please add this suggestion on the speclite
14:30:02 <jokke_> simple way to indicate that we don't want to patch it ... if we're looping anyways doing it in patches is just tons of extra queries into the db that are not needed
14:30:14 <jokke_> abhishekk: sure
14:30:29 <abhishekk> any specific reason for not doing it Mitya_Eremeev
14:30:57 <Mitya_Eremeev> sometimes db can be very huge and sometimes it's better to divide deleting in steps
14:31:20 <dansmith> in nova we have a date-based filter as well, so you can say "older than last month"
14:31:23 <Mitya_Eremeev> so max-rows helps regarding performance
14:31:48 <abhishekk> dansmith, I think similar we have age-in-days option
14:32:08 <dansmith> okay cool, I'm not sure why anyone would use max-rows if they have an age-based option
14:32:35 <jokke_> dansmith: well we for some reason default that to 100 rows
14:32:36 <Mitya_Eremeev> max-rows is just for db performance
14:32:54 <dansmith> jokke_: oh max-rows counts against an age-based purge as well?
14:33:15 <Mitya_Eremeev> yes
14:33:22 <dansmith> yeah, that seems weird to me
14:33:29 <jokke_> dansmith: yeah yeah ... by default it does only 100 rows and you loop it as needed or give it sufficient value
14:33:33 <jokke_> makes no sense
14:33:36 <dansmith> maybe loop and do $max_rows at a time until there are no more that satisfy the age query?
14:33:45 <dansmith> I mean, make the tool do that for them
14:33:48 <abhishekk> I think if age-in-days is mentioned then it should ignore max-rows
14:33:53 <jokke_> as it just introduces more load for db to loop the data and identify those rows
14:34:59 <abhishekk> I think its time to revisit glance-manage utility tool and do some enhancements there as well
14:35:16 <dansmith> abhishekk: well, depends on how the loop works - but max-rows may be desirable to avoid loading a billion things into memory just to delete them,
14:35:28 <dansmith> but the user should expect it to behave sanely, regardless of internal batching, IMHO
14:35:35 <jokke_> abhishekk: they both have default values
14:36:17 <jokke_> dansmith: but that should be handled in the db_api side, rather than from the command client
14:36:23 <abhishekk> hmm, I need to recheck how it works, we never looked back once we added purge image table separately
14:36:26 <jokke_> batching I mean
14:36:29 <jokke_> if needed
14:36:41 <dansmith> jokke_: as long as it's not handled by the user in the shell, I agree :)
14:36:55 <abhishekk> :D
14:37:12 <jokke_> but any case the 100 line limit is ridiculous
14:37:23 <abhishekk> We can continue this discussion on spec lite
14:37:26 <jokke_> mhm
14:37:44 <abhishekk> and if required we can change it to spec where we can target all the improvements at once
14:37:59 <abhishekk> Mitya_Eremeev, thank you for bringing this up
14:38:09 <Mitya_Eremeev> Thank you all !
14:38:19 <abhishekk> moving ahead
14:38:39 <abhishekk> #topic in-flight encryption without Barbican Consumer API
14:38:46 <abhishekk> rosmaita, this is you
14:39:06 <rosmaita> yeah, my -1 on that may be preventing people from looking at it
14:39:17 <abhishekk> I think I have added in PTG etherpad some feedback but forget to add it to spec before going on Vacation
14:39:21 <rosmaita> -1 because it's not targeted for yoga
14:40:02 <abhishekk> will again have a look after the meeting or early tomorrow
14:40:03 <rosmaita> but i was wondering if dansmith and jokke_ still have reservations about doing this without the secret consumer API?
14:40:48 <abhishekk> Only reservation is whether we should allow commented code as place holder or we should have dependent patch instead of having commented code in code base
14:41:09 <dansmith> -1 on commented code, but I'm not up to date on the current state
14:41:24 <rosmaita> i agree with dansmith on that
14:42:01 <abhishekk> Yeah, its better to have dependent patch
14:42:12 <rosmaita> what is it going to depend on?
14:42:40 <jokke_> rosmaita: as actual stable feature yes, I've been promoting doing this as experimental so the development work can move on while waiting for the consumer api for over cycle now ;)
14:43:22 <rosmaita> ok, my only interest is that i want them to get this thing to the state where they can run end-to-end CI tests
14:43:23 <jokke_> rosmaita: the non-exiting release patch of barbican to release the consumer api ;)
14:43:35 <abhishekk> https://review.opendev.org/c/openstack/glance/+/705445/4/glance/api/v2/images.py
14:43:37 <rosmaita> because there are a bunch of edge cases that will have to be dealt with, i suspect
14:43:51 <abhishekk> rosmaita, we can move commented code from this patch to a separate patch
14:44:21 <rosmaita> i wonder about that patch
14:44:27 <rosmaita> the comment i mean
14:44:41 <rosmaita> i thought castellan was a backend-neutral library
14:44:50 <rosmaita> so unlikely to have support for this thing
14:45:11 <rosmaita> it doesn't even support naming a secret (unless the version i use is way out of date)
14:45:47 <rosmaita> that's a separate issue, i guess
14:46:00 <abhishekk> right
14:46:33 <abhishekk> is encryption weekly meeting still going on?
14:46:53 <rosmaita> yes, though i keep missing it because i am stupid about the time change
14:47:17 <rosmaita> mondays at 1300 UTC, i think
14:47:18 <abhishekk> let me know the timing I will join it (next time)
14:47:26 <abhishekk> I think it is on Monday
14:47:37 <abhishekk> Ok, will try to be there this time
14:47:56 <rosmaita> yes, 1300 utc in #openstack-meeting
14:47:56 <abhishekk> moving ahead
14:48:00 <rosmaita> thanks!
14:48:03 <abhishekk> cool, thank you
14:48:19 <abhishekk> #topic Open discussion
14:48:44 <abhishekk> I need to revisit PTG discussions to start up pending works
14:48:56 <abhishekk> I guess Cache API is one of them
14:49:21 <abhishekk> So in next meeting we will have some targets for Milestone 2
14:49:37 <abhishekk> that's it from me
14:49:49 <abhishekk> Kindly put some time in spec reviews
14:51:40 <abhishekk> anything else ?
14:52:11 <jokke_> not from me
14:53:27 <rosmaita> not me
14:53:56 <dansmith> nay
14:54:24 <abhishekk> cool, lets wrap up for the day
14:54:26 <abhishekk> thank you all
14:54:34 <abhishekk> have a great week ahead
14:55:05 <jokke_> thanks
14:55:23 <abhishekk> #endmeeting