14:00:15 <abhishekk> #startmeeting glance
14:00:27 <openstack> Meeting started Thu Jan 23 14:00:15 2020 UTC and is due to finish in 60 minutes.  The chair is abhishekk. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:00:28 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:00:29 <abhishekk> #topic roll call
14:00:30 <openstack> The meeting name has been set to 'glance'
14:00:35 <jokke_> o/
14:00:36 <abhishekk> #link https://etherpad.openstack.org/p/glance-team-meeting-agenda
14:00:38 <rosmaita> o/
14:00:40 <abhishekk> o/
14:00:44 <yebinama_> o/
14:00:53 <abhishekk> Cool
14:00:57 <abhishekk> lets start
14:01:25 <abhishekk> #topic release/periodic jobs update
14:01:58 <abhishekk> We still have 1 failure in periodic job which is causing because of oslo_config new version
14:02:12 <abhishekk> #link https://review.opendev.org/701580
14:02:31 <abhishekk> I have added workaround on this patch but there was no response from the owner
14:03:04 <abhishekk> So workaround sounds good to you people I will go ahead and submit this patch
14:03:45 <abhishekk> Kindly comment on the patch about the same, moving ahead
14:03:54 <abhishekk> Ussuri milestone 2 - 4 weeks to go
14:04:09 <abhishekk> We have 3 specs targeted to milestone 2
14:04:47 <abhishekk> So kindly focus on reviews, I don't want to miss this deadline
14:05:27 <abhishekk> This Monday we have released stable train and stein version
14:05:56 <abhishekk> any comments so far?
14:06:16 <abhishekk> cool
14:06:20 <abhishekk> #topic Multiple store import plugins
14:06:35 <abhishekk> multiple import impl
14:06:46 <abhishekk> yebinama_, you want to talk about this?
14:06:58 <yebinama_> I've added the notification part
14:07:06 <abhishekk> I guess except documentation and releasenote everything is covered
14:07:17 <abhishekk> yebinama_, yes I saw that, looks good to me
14:07:28 <jokke_> Are the specs finally green?
14:07:32 <yebinama_> Yep only these are missing
14:07:33 <abhishekk> #link https://review.opendev.org/667132
14:07:40 <abhishekk> jokke_, yes
14:08:17 <abhishekk> jokke_, rosmaita kindly review the patch ^^^
14:08:23 <yebinama_> Should I create a new doc.rst for multi store import?
14:08:50 <abhishekk> yebinama_, I guess you could add in existing file only
14:09:01 <jokke_> ++
14:09:07 <abhishekk> I have added info about copying stuff it in existing file
14:09:13 <yebinama_> Ok fine
14:09:13 <jokke_> for adding the details to the existing import docs
14:09:16 <abhishekk> refer my patch for the same
14:09:28 <yebinama_> Ok thanks
14:09:29 <abhishekk> yebinama_, you also need to modify api-ref guide as well
14:09:55 <abhishekk> Copying existing image impl
14:10:06 <abhishekk> #link https://review.opendev.org/696457
14:10:20 <abhishekk> So regarding this patch only release note is remaining
14:10:35 <abhishekk> I have added functional test to cover the revert logic as well
14:10:52 <yebinama_> abhishekk: I'll work on addind this
14:11:20 <abhishekk> with this test we are able to verify that on failure we are deleting image data from only those stores where it was copied and not from the existing stores
14:11:31 <abhishekk> yebinama_, ack
14:11:41 <rosmaita> that sounds like a good test!
14:11:58 <abhishekk> Hopefully
14:12:06 <yebinama_> Great I'll take a look
14:12:32 <abhishekk> I have written two test for these scenario with considering 'all_stores_must_succeed' parameter
14:13:16 <abhishekk> As copying patch is dependent on yebinama_'s import patch, we need to first put priority on reviewing and merging his patch
14:13:50 <abhishekk> I have tested his patch functionally with 3 file and one ceph backend
14:14:02 <yebinama_> And just to let you know I'll be on holidays starting 5 of february
14:14:03 <rosmaita> please remove "WIP" and "PoC' from the commit messages on those patches, that's why i have been ignoring them
14:14:19 <abhishekk> rosmaita, ack
14:14:48 <abhishekk> yebinama_, so we need to make sure we can complete all the work before that and get this merged
14:14:52 <yebinama_> rosmaita: ok i'll update the commit
14:14:59 <rosmaita> ty
14:15:13 <abhishekk> In case if some review comments come then should I fix those on your behalf?
14:15:34 <yebinama_> abhishekk: I'll try to bring my laptop so I should b3 able to work on it
14:15:54 <abhishekk> yebinama_, cool
14:16:20 <abhishekk> jokke_, rosmaita need your expert opinions on the patches
14:16:31 <abhishekk> Moving ahead
14:16:42 <abhishekk> #topic Delete image from single store
14:17:04 <jokke_> We can skip this topic and move on
14:17:12 <abhishekk> I saw jokke_ has abandoned the patch, any particular reason
14:18:08 <abhishekk> jokke_, at some point we need to work on this
14:18:39 <jokke_> It was clear to me that we won't be getting into agreement between what rosmaita wanted nd how far I was willing to push the consistencies with the rest of our API
14:19:23 <rosmaita> all i asked for was to fix some typos!
14:19:46 <abhishekk> so what if instead of using location's pop method we use store delete method to delete the location?
14:20:00 <abhishekk> this will avoid race condition and all the fuss
14:20:46 <abhishekk> as we have store identifier in location metadata and we have store instance in memory, we can retrieve the same and call store delete method directly
14:20:58 <rosmaita> somehow the location needs to be deleted, and i believe location.remove() calls location.pop()
14:21:11 <jokke_> mhm
14:21:21 <rosmaita> there is some really weird coupling in the code
14:21:33 <jokke_> We got the race condition solved that wasn't the problem left
14:21:56 <abhishekk> jokke_, I really don't want you to throw away your efforts
14:22:24 <rosmaita> let me say something here about the consistency issue and how we talk about stuff in docs and error messages
14:22:37 <rosmaita> before multistore, the locations stuff was esoteric
14:22:48 <rosmaita> now, with the edge stuff we need to be very clear
14:22:49 <abhishekk> We can fix the typos and if required later revisit all multi-store related work to reword the help text
14:22:59 <abhishekk> even I have made mess at some places
14:23:04 <rosmaita> that "Image" and "image data" are different
14:23:16 <rosmaita> before, you could talk about them as pretty much the same
14:23:33 <rosmaita> but now we have the Image record and various stores where the image data is stored
14:23:47 <rosmaita> so we need to be really clear in talking about what's being removed/deleted
14:24:17 <rosmaita> that's why i'm insisting on the message changes about image data is being removed from store
14:24:57 <rosmaita> i think the locations are an implementation detail
14:25:06 <rosmaita> so we don't need to discuss them
14:25:08 <abhishekk> jokke_, please reconsider and restore the patch, its required otherwise we need to be depend on location API to remove the location
14:26:05 <abhishekk> Moving ahead
14:26:13 <abhishekk> #topic Open Discussion
14:26:29 <abhishekk> So we need to focus on these three patches
14:26:59 <abhishekk> jokke_, rosmaita, smcginnis kindly review these patches
14:27:37 <alistarle> Hi guy's, just a question about implementation of limits/quota in glance
14:27:53 <abhishekk> alistarle, shoot
14:28:11 <smcginnis> abhishekk: Which three patches need attention? The above topics?
14:28:26 <alistarle> We experience some capacity planning issue for image, so we wanted to implement some quota out there, we see that there is only a conf option about that
14:28:34 <abhishekk> smcginnis, yes topic 2 and 3 from the agenda
14:28:54 <rosmaita> alistarle: yes, our "quotas" are really just hard limits
14:29:01 <alistarle> And I see that Keystone implement the unified limit framework accross all openstack project, did you look at that ?
14:29:19 <smcginnis> Will do.
14:29:24 <alistarle> Seems that only nova are implementing it now, but maybe it is the time to implement it in glance no ?
14:29:25 <rosmaita> alistarle: we would be happy to review your code if you want to implement that
14:29:38 <abhishekk> smcginnis, thank you
14:30:22 <alistarle> Yes I would like to work on that, seems the library oslo.limits must be used, but I don't understand some point of their framework
14:30:23 <abhishekk> alistarle, also it will be great if you submit the glance-specs for the same
14:30:31 <alistarle> Sure
14:30:51 <smcginnis> alistarle: I'm not sure if the unified limits work is complete. The last I read on it, nova was a first test case to help flush out the API. If it's ready now, it would be great to get some details about how to implement its use.
14:30:56 <abhishekk> it will help us to understand the design and providing suggestions
14:31:19 <jokke_> alistarle: I have not looked into it. Is that now something that the majority of OpenStack community finally got behind or is that something implemented in Keystone when the consensus was not reached? ;)
14:31:34 <alistarle> Yes that's what I was thinking, because it respond very well for enforcing /consulting limits, but not a all about how a user can consult it's own usage
14:32:52 <alistarle> The point is we needed quotas, so I think working on unified limits is better than trying to do something on our side
14:34:24 <smcginnis> Definitely agree it's about time we get to a common solution for this.
14:34:51 <abhishekk> +1
14:35:13 <jokke_> alistarle: happy to hear that there is someone like you actually needing it and willing to work to make it happen. This has been topic that comes up couple of times per year for quite a few years and no-one has been willing to step up and tell that they need it enough to start driving the efforts
14:35:38 <abhishekk> ++
14:37:17 <alistarle> Good, I will try to make by best so ;)
14:37:27 <jokke_> :)
14:38:08 <abhishekk> smcginnis, regarding oslo_config related failure, I will report a bug and submit a patch for the same
14:39:01 <smcginnis> Great, thanks abhishekk!
14:39:03 <abhishekk> or will update the same patch
14:39:36 <abhishekk> anything else to discuss?
14:39:49 <abhishekk> jokke_, rosmaita, smcginnis, yebinama_, alistarle ?
14:40:04 <jokke_> abhishekk: I'm pretty sure Stephen doesn't mind if you take that patch over with better approach for us
14:40:06 <smcginnis> Nothing from me.
14:40:15 <jokke_> nothing from me
14:40:22 <abhishekk> jokke_, I will add him as co-author
14:40:51 <rosmaita> i may have missed something
14:41:03 <rosmaita> did jokke_ say whether he will un-abandon that patch?
14:42:08 <abhishekk> he has not, but he will definitely consider as we need that patch
14:42:44 <rosmaita> ok
14:42:58 <jokke_> No I did not ... will see how much more time I have to keep fighting with that
14:44:49 <abhishekk> Ok, lets windup to utilize remaining 15 minutes
14:44:57 <abhishekk> Thank you all
14:45:07 <yebinama_> Bye
14:45:16 <rosmaita> bye!
14:45:46 <smcginnis> Thanks!
14:45:48 <jokke_> thanks
14:45:54 <abhishekk> #endmeeting