14:01:38 <abhishekk> #startmeeting glance
14:01:39 <openstack> Meeting started Thu Aug  6 14:01:38 2020 UTC and is due to finish in 60 minutes.  The chair is abhishekk. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:01:40 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:01:42 <openstack> The meeting name has been set to 'glance'
14:01:45 <abhishekk> #topic roll call
14:01:48 <jokke> o/
14:01:50 <abhishekk> #link https://etherpad.openstack.org/p/glance-team-meeting-agenda
14:01:52 <abhishekk> o/
14:01:53 <mhen> o/
14:01:59 <dansmith> o/
14:02:13 <Luzi> o/
14:02:21 <abhishekk> lets wait couple of mins for rosmaita
14:03:20 <rosmaita> o/
14:03:30 <abhishekk> cool, lets start
14:03:31 <rosmaita> (i am only partially here)
14:03:34 <abhishekk> #topic Updates
14:03:36 <Steap> o/
14:03:46 <Steap> rosmaita: that is fine, we only need your brain!
14:04:11 <abhishekk> Welcome Dan to the cores group
14:04:13 <rosmaita> there is not enough of that to go around
14:04:22 <rosmaita> congratulations, Dan!
14:04:31 <dansmith> woo
14:04:32 <Steap> gg, Dan!
14:04:43 <abhishekk> Welcome and congratulations Dan!!
14:04:51 <abhishekk> Steap, means cyril?
14:05:04 <Steap> oh
14:05:17 <Steap> "gg" is "good game", but it may not be completely mainstream yet
14:05:20 * Steap is ahead of his time
14:05:23 <rosmaita> Steap: you need to call yourself 'Bruce', it will be less confusing
14:05:34 <abhishekk> Sorry :D
14:06:04 <abhishekk> Dan is helping us a lot and hopefully he will stays with us little longer :P
14:06:11 <abhishekk> Moving ahead
14:06:21 <abhishekk> #topic release/periodic jobs update
14:06:34 <abhishekk> stable/train and stable/ussuri release for glance
14:07:08 <abhishekk> Still there are couple of patches under review or in the gate, So I will suggest we should release both on Monday
14:07:13 <abhishekk> any objections?
14:07:29 <abhishekk> jokke ^^
14:07:48 <jokke> if we get the ussuri backports in I can push the release patch up to review today
14:08:07 <abhishekk> If that happens then its good
14:08:16 <abhishekk> otherwise we will release on Monday
14:08:28 <jokke> client patch is up already and has +2 from smcginnis the glance side needs those backports still
14:08:51 <abhishekk> yeah
14:08:55 <abhishekk> Final release for non clients libraries - 3 weeks from now
14:09:17 <abhishekk> that means we need to get cinder multiple stores specs in by tomorrow EOD
14:09:37 <abhishekk> I am ready with the PoC just need to work on tests and polishing little bit
14:10:00 <abhishekk> rosmaita, smcginnis kindly have a look at cinder multiple stores specs
14:10:15 <rosmaita> ack
14:10:22 <abhishekk> V3 milestone 4 weeks from now
14:10:30 <abhishekk> just month remaining
14:11:02 <abhishekk> we need to get image encryption, sparse image handling, virtual size calculation and multiple cinder stores work done before this
14:11:17 <abhishekk> I guess Luzi is here
14:11:37 <abhishekk> Could you please share updates about image encryption work?
14:11:46 <abhishekk> mhen, as well
14:12:44 <abhishekk> ok, we will get back to this in Open discussion
14:12:48 <Luzi> we are currently waiting on the Barbican Secret Consumer API
14:13:02 <Luzi> they had some issues with the gate-jobs
14:13:07 <abhishekk> what is ETA for that
14:13:21 <abhishekk> are we positive to get this done in Victoria?
14:13:54 <Luzi> i am not sure
14:14:32 <abhishekk> Ok
14:14:43 <abhishekk> thank you for the update
14:15:06 <abhishekk> Keep us posted about the progress if possible
14:15:09 <mhen> mostly depends on Barbican and how quick their new API is available to us and how much time we have left for the necessary adjustments to incorporate the new API then
14:15:49 <abhishekk> mhen, I am asking because if we are certain then we can file for FFE if required
14:16:40 <abhishekk> Lets revisit the status in next week and then we will decide whether to move it to next cycle or not
14:16:55 <abhishekk> Moving ahead
14:17:07 <abhishekk> #topic Specs status
14:17:40 <abhishekk> we still have 5 specs open for reviews
14:17:45 <abhishekk> Make cinder driver compatible with multiple stores - https://review.opendev.org/695152 - Need reviewes
14:17:45 <abhishekk> Introspect import plugin to calculate virtual size of image - https://review.opendev.org/741121 - Need reviews
14:17:46 <abhishekk> Optimize Ceph store network usage - https://review.opendev.org/#/c/740980/ - Need reviews
14:17:46 <abhishekk> Update proposal for duplication image download - https://review.opendev.org/734683 - Need reviews
14:17:47 <abhishekk> Cache API - https://review.opendev.org/#/c/665258/ - Need reviews
14:18:07 <abhishekk> Out of which 1st are almost ready to go
14:18:15 <abhishekk> Code is ready for both of them
14:18:26 * Steap really needs to work on the cache api spec a bit more
14:18:28 <abhishekk> SO kindly review them on top priority
14:18:46 <alistarle> For https://review.opendev.org/#/c/740980/, I still think example will help to validate the optimization
14:19:03 <dansmith> abhishekk: I will +2 the virtual size one after this when I re-read it, but should be fine I think
14:19:14 <alistarle> jokke I can try to find it if you want (but I have also lot of work sparse upload side :p  )
14:19:15 <abhishekk> dansmith, ack
14:19:36 <abhishekk> alistarle, focus on sparse upload first
14:19:48 <abhishekk> :D
14:20:09 <abhishekk> Moving to next topic
14:20:16 <abhishekk> #topic Work moving to W cycle
14:20:29 <jokke> alistarle: yeah, lets try to get your sparse stuff in, that rest of the ceph stuff has been kind of waiting to layer on top of the work you guys already did
14:21:05 <abhishekk> Due to lack of time I am suggesting to move below items to next cycle
14:21:14 <abhishekk> Code cleanup (ToDo items from code) - https://ethercalc.openstack.org/kq4gwn9gqzck
14:21:29 <abhishekk> restrict Duplicate downloads
14:21:35 <abhishekk> Cache API
14:21:41 <abhishekk> Cluster awareness for glance
14:21:48 <abhishekk> Remove single store configuration
14:22:00 * Steap is OK with moving Cache API
14:22:25 <abhishekk> I am still putting efforts on removing single store configuration but I am doubtful to complete it within 3 weeks
14:22:33 <jokke> I was just gonna ask if Steap thinks the cache api is doable within the timeframe we have
14:22:55 <abhishekk> For Cache API I have PoC ready
14:23:27 <jokke> I'm around for about two and half weeks still. Then i will disappear into the forest for a week :P
14:23:27 <abhishekk> https://review.opendev.org/#/c/672876/
14:24:01 <abhishekk> I hope everyone agrees on the work I selected to Move to next cycle
14:24:45 * abhishekk good for you jokke :D
14:25:39 <abhishekk> Cool, moving ahead to Open discussion
14:25:54 <abhishekk> #topic Open discussion
14:26:18 <abhishekk> Between Luzi and mhen thank you for joining in and sharing the updates
14:26:30 <Steap> jokke: yeah I still need to disappear as well, so... :)
14:27:06 <abhishekk> I have one topic for discussion
14:27:19 <abhishekk> I have seen code-coverage job for most of the core projects
14:27:31 <abhishekk> like nova, cinder, neutron etc
14:27:46 <abhishekk> any idea what is usefulness of it?
14:27:56 <abhishekk> rosmaita, dansmith ^^
14:28:17 <jokke> We used to have one I think we got rid of it as no-one ever paid any attention to it
14:28:28 <dansmith> in nova, it's not much use, IMHO
14:29:07 <dansmith> in other projects I've used a coverage tracking tool to make sure that stuff that is added gets tested,
14:29:10 <rosmaita> well, it's useful to see what the coverage is on new patches
14:29:10 <abhishekk> as far as I think the purpose should be look at the results and see whether code added is covered or not
14:29:24 <dansmith> but with such a large and indirectly-connected project, it can be hard
14:29:33 <abhishekk> do we really have time to check this?
14:29:37 <jokke> yeah, it was one of those things the Rackspace/Intel guys were active on and when that project got carpet pulled under it no-one just paid any attention to it so IIRC we removed the job to release that bit of resources back to infra pool
14:30:15 <dansmith> glance seems to be missing a lot more coverage than nova does though, so it may be useful here if the data is presented in a useful format
14:30:29 <dansmith> but it requires diligence by the cores to check it and force people to write tests
14:30:37 <abhishekk> ++
14:30:41 <dansmith> so if that's not going to happen, then it's just a waste of infra resources
14:30:58 <jokke> i think tox gives that info for you, so you can just run it locally to see what's in the report
14:31:06 <dansmith> it's also something you can do by downloading a patch and determining it yourself
14:31:19 <rosmaita> which nobody does
14:31:21 <dansmith> tox doesn't unless you run pycoverage or something, that I know of,
14:31:28 <dansmith> but yes, it's doable locally
14:31:52 <abhishekk> I guess next cycle we should add one goal to increase code coverage of glance
14:32:05 <jokke> we have tox env called cover for it
14:32:22 <abhishekk> I think we have tox -e coverage or cover in glance tox.ini
14:32:25 <dansmith> jokke: yeah, that's not tox, that's just us configuring tox to run a coverage tool
14:32:48 <dansmith> abhishekk: just to be clear, adding a job does not increase coverage... only humans increase coverage :)
14:32:50 <jokke> dansmith: sure yeah, but I mean we do have tooling via tox to do it locally was what I meant
14:33:02 <jokke> dansmith: ++
14:33:03 <abhishekk> dansmith, yeah I know that :D
14:33:33 <dansmith> abhishekk: just saying, it's easy to say "we'll add the job" and start consuming resources, but then never do anything about it, so... need some policies and procedures... and policing :)
14:33:53 <jokke> dansmith: and that was why we removed the job last time :D
14:34:08 <abhishekk> dansmith, yes, thats why I asked what are the benefits of having it
14:34:17 <dansmith> yep, but if the policing will be there, I'm all for it, because.. there's a lot of uncovered stuff
14:34:25 <alistarle> can't the job be a voting one then ?
14:34:34 <alistarle> we can have a gate job only for new code coverage
14:34:41 <dansmith> alistarle: not really,
14:35:13 <dansmith> because there are always lines you can't cover that don't matter much, and making the job refuse to merge a patch because you didn't roll over some docstring line is a real problem
14:35:25 <abhishekk> we need something like the job should fail if that patch has less than 50% code coverage
14:35:25 <rosmaita> in cinder, we decided at the ptg to do it as a non-voting job and revisit at the next ptg to see if anyone is using it
14:35:26 <dansmith> the tool generally just *reports* coverage, and you have to decide whether you're okay with the report
14:36:10 <alistarle> we can also restrict the coverage on some part of the code
14:36:11 <abhishekk> yeah, that's really time consuming though it is good practice
14:36:39 <jokke> also it was driving behaviour where people wrote pretty much literally tests like assetEqual(configoption, configdefault) just to get the coverage number up. We need meaningful tests, not just more tests
14:36:59 <abhishekk> ++
14:37:47 <alistarle> Yep, and btw I think we are more lacking of corner cases integration test than unit test
14:39:03 <dansmith> we could do a lot better by increasing the requirements the reviewers place on new patches and that's "free" in that it only demands the core time part of the coverage equation
14:39:18 <dansmith> and then worry about tooling when we're starting to wonder how close to 100% we are :)
14:39:24 <jokke> ++
14:39:39 <abhishekk> :D
14:41:37 <abhishekk> Ok, any other topic for discussion??
14:41:54 <alistarle> Concerning sparse upload, we finally admit it will be far easier to test all the case from gerrit than on premise :p  So we submit this first version : https://review.opendev.org/#/c/744282/
14:42:17 <jokke> rosmaita: smcginnis: the two ussuri patches we're looking to include in the stable release are https://review.opendev.org/744997 and https://review.opendev.org/745100 if you have time
14:42:41 <alistarle> We still have issue with covering all checksum validation case, but we are working really faster in that condition than in our own environment
14:43:04 <abhishekk> alistarle, let us know if you need anything
14:43:26 <alistarle> I hope next week it will be fully fonctionnal
14:43:47 <abhishekk> cool
14:43:55 <alistarle> and I also submit a talk to the virtual summit about our multi AZ setup, will talk about glance multi store for sure
14:44:12 <abhishekk> \o/
14:44:38 <jokke> nice
14:45:01 <alistarle> But I don't know if it will be selected ;)
14:45:12 <abhishekk> :d
14:45:27 <abhishekk> B+ve
14:46:31 <abhishekk> Ok that's it for me today
14:47:17 <abhishekk> lets wrap up unless anyone wants to discuss or share any updates
14:48:40 <jokke> I have nothing else for now
14:48:50 <abhishekk> cool, thank you all
14:48:55 <abhishekk> have a nice time ahead
14:49:10 <jokke> thanks all
14:49:47 <abhishekk> #endmeeting