14:00:29 <pdeore> #startmeeting glance 14:00:29 <opendevmeet> Meeting started Thu May 30 14:00:29 2024 UTC and is due to finish in 60 minutes. The chair is pdeore. Information about MeetBot at http://wiki.debian.org/MeetBot. 14:00:29 <opendevmeet> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 14:00:29 <opendevmeet> The meeting name has been set to 'glance' 14:00:29 <pdeore> #topic roll call 14:00:29 <pdeore> #link https://etherpad.openstack.org/p/glance-team-meeting-agenda 14:00:36 <pdeore> o/ 14:01:02 <dansmith> o/ 14:01:45 <pdeore> Brian, Cyril & Mridula are not joining the meeting 14:02:11 <pdeore> abhishekk, you around ? 14:02:38 <abhishekk> o/ in another meeting though 14:02:56 <pdeore> ack, let's get started then 14:03:05 <pdeore> #topic release/periodic jobs updates 14:03:16 <pdeore> M2 is 5 weeks from now 14:03:42 <pdeore> periodic jobs: all tips jobs are still failing since this patch is not yet merged 14:03:45 <pdeore> #link https://review.opendev.org/c/openstack/glance/+/919817 14:04:17 <pdeore> it's depends on this glance-tempest-plugin patch #link https://review.opendev.org/c/openstack/glance-tempest-plugin/+/916332 14:04:26 <abhishekk> will have a look later 14:04:28 <pdeore> kindly please have a look 14:04:31 <pdeore> ack, Thanks 14:04:37 <pdeore> moving ahead 14:04:47 <pdeore> #topic Important Specs to Review 14:05:06 <pdeore> deprecate metadata-encryption-key - #link https://review.opendev.org/c/openstack/glance-specs/+/916178 14:05:07 <pdeore> Revised spec for Image Encryption - #link https://review.opendev.org/c/openstack/glance-specs/+/915726 14:05:44 <pdeore> please have a look 14:06:12 <pdeore> moving to next 14:06:14 <pdeore> #topic swift store issue 14:06:26 <pdeore> #link https://bugs.launchpad.net/glance/+bug/2061947 14:06:43 <pdeore> So when swift store is enabled with swift_store_multi_tenant and swift_store_config_file together, glance stores-info api fails with 500 14:07:06 <pdeore> so here there are 2 issues and I submitted a fix as well 14:07:16 <pdeore> #link https://review.opendev.org/c/openstack/glance/+/920170 14:07:35 <pdeore> I have doubt about fix for the second issue 14:08:08 <pdeore> When swift multitenant enabled, container is not getting created during configuration but during image create it's getting created if it's not set during store configuration. 14:08:35 <pdeore> So is it ok to return container value None if it's not set while fetching the store details or this should be fixed at store side? 14:08:39 <abhishekk> what do you mean by during configuration? 14:09:09 <pdeore> during configuring swift store 14:09:29 <pdeore> incase of single tenant it gets set 14:09:31 <abhishekk> you mean at service start when store is initialized? 14:09:42 <pdeore> yeah 14:09:54 <abhishekk> I think that needs to be fixed at store side 14:10:35 <abhishekk> if it's set for single store then it should be same for multi tenant imo 14:10:40 <pdeore> yeah because the container value could be either None or glance_* 14:11:06 <abhishekk> ack 14:11:26 <pdeore> ack, Thanks 14:11:40 <pdeore> that's it from me 14:11:47 <pdeore> let's move to open discussions 14:12:07 <pdeore> ohh sry, one more thing 14:12:38 <pdeore> There is already a note added regarding the swift configuration for multitenant in the doc 14:12:48 <pdeore> #link https://docs.openstack.org/glance/latest/configuration/configuring.html#configuring-swift-configuration-file 14:13:25 <pdeore> so do we need some more details or this is enough? 14:13:51 <abhishekk> we need validation around that 14:14:18 <pdeore> ok 14:14:46 <pdeore> Thanks 14:14:57 <pdeore> moving to open discussions 14:15:04 <pdeore> #topic Open Discussions 14:15:27 <abhishekk> my NFS spec has two +2s so ideally you can approve that unless have any objection 14:16:11 <pdeore> yeah will do that 14:16:15 <abhishekk> also I have submitted patch in oslo_middleware, please have a look at that 14:16:55 <abhishekk> #link https://review.opendev.org/c/openstack/oslo.middleware/+/920055?usp=search 14:17:15 <pdeore> ack 14:17:45 <pdeore> Regarding hash calculation command line utility, there is suggestion from abhishekk to have a separate command line utility instead of adding the new command under glance-manage to avoid complexity 14:17:57 <pdeore> I would like to hear others opinion on the same 14:18:38 <pdeore> cyril is suggesting not to have the separate one 14:18:45 <pdeore> dansmith, what would you suggest? 14:19:10 <dansmith> why a separate command? 14:19:14 <abhishekk> So related to that, since we calculating hash for the snapshot, can we also set image size for the same? 14:19:58 <abhishekk> I was suggesting to have different command line tool same as glance-cache-prefetcher to keep it different 14:20:26 <dansmith> but why? 14:20:44 <abhishekk> glance-manage image calculate-hash doesn't suited well for me 14:21:02 <abhishekk> we have glance manage for db and metadef loading related work only 14:21:44 <dansmith> this is for re-calculating the hash on an existing image that doesn't have one right? 14:21:50 <abhishekk> yes 14:22:00 <abhishekk> i.e for legacy sanpshots 14:22:07 <dansmith> is there some reason that shouldn't actually be an api activity? 14:22:15 <dansmith> I mean, it's probably *too* high level even for glance-manage, 14:22:32 <abhishekk> yyou mean periodic job or at the start of the service? 14:22:40 <dansmith> because you might delegate that activity to some person in your org or a script with some limited credentials 14:22:58 <dansmith> no, I mean trigger it via the api and let it spawn a thread to do it just like when you're uploading 14:23:11 <abhishekk> or separate API altogether 14:23:28 <abhishekk> ack, this will be more easy approach 14:24:25 <abhishekk> pdeore: ^^ 14:24:27 <dansmith> I dunno about easy to implement, but definitely easier for the operators I think, and avoids the need for another dedicated tool 14:24:37 <pdeore> so if we have a separate api then do we need a spec for this? 14:24:40 <dansmith> and matches api-based image prefetching like we have now 14:24:57 <abhishekk> agree 14:25:18 <pdeore> ok 14:25:35 <abhishekk> yes, you can submit a spec for this imo 14:26:20 <abhishekk> also coming back to my suggestion, since we calculating hash for the snapshot, should we also compute and set image size ? 14:26:20 <pdeore> ack 14:26:32 <dansmith> abhishekk: definitely 14:26:49 <abhishekk> this will also solve our zero size image case 14:27:36 <pdeore> sure, I will do that too 14:27:53 <abhishekk> sorry for late suggestion, but I think it's hust couple of lines addition to the code and test 14:28:16 <abhishekk> thank you! I am done! 14:28:23 <pdeore> :) 14:28:54 <pdeore> I think this should be separate out from the location import chain or may be would be btr on top of it 14:29:06 <pdeore> *separated 14:29:11 <abhishekk> size calculation? 14:29:21 <pdeore> no hash calculation 14:29:44 <abhishekk> new API, you can submit that as a separate work 14:30:05 <pdeore> ok 14:30:09 <pdeore> Thanks ! 14:30:21 <pdeore> that's it from me 14:30:41 <pdeore> anyone has anything else to highlight? or we can wrap up early? 14:31:49 <pdeore> seems nothing, let's wrap up then 14:32:00 <abhishekk> thank you! 14:32:01 <pdeore> Thanks for joining !! 14:32:06 <pdeore> #endmeeting