07:00:30 <yoctozepto> #startmeeting masakari
07:00:31 <openstack> Meeting started Tue Feb  9 07:00:30 2021 UTC and is due to finish in 60 minutes.  The chair is yoctozepto. Information about MeetBot at http://wiki.debian.org/MeetBot.
07:00:32 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
07:00:34 <openstack> The meeting name has been set to 'masakari'
07:00:37 <yoctozepto> #topic Roll-call
07:00:41 <yoctozepto> \o/
07:01:39 <jopdorp> o/
07:02:18 <yoctozepto> hi jopdorp
07:02:35 <yoctozepto> suzhengwei is on holidays (lunar new year celebration) so let's start
07:02:39 <yoctozepto> #topic Agenda
07:02:51 <yoctozepto> * Roll-call
07:02:51 <yoctozepto> * Agenda
07:02:51 <yoctozepto> * Announcements
07:02:51 <yoctozepto> * Review action items from the last meeting
07:02:51 <yoctozepto> * CI status
07:02:51 <yoctozepto> * Backports pending reviews
07:02:52 <yoctozepto> * Release planning https://etherpad.opendev.org/p/masakari-wallaby-vptg
07:02:52 <yoctozepto> * Open discussion
07:03:06 <yoctozepto> #topic Announcements
07:03:11 <yoctozepto> I have none
07:03:23 <jopdorp> Me neither
07:03:43 <yoctozepto> we are at R-9
07:04:00 <yoctozepto> so it's very very close to freeze and release
07:04:16 <yoctozepto> in fact feature freeze and the 3rd milestone is in 4 weeks
07:04:26 <yoctozepto> need to merge some features!
07:04:26 <jopdorp> Ok
07:04:33 <yoctozepto> #topic Review action items from the last meeting
07:04:37 <yoctozepto> there were none
07:04:42 <yoctozepto> #topic CI status
07:04:56 <yoctozepto> green now
07:05:14 <yoctozepto> we were hit by alembic upgrade that caused taskflow to fail
07:05:21 <yoctozepto> but that was quickly handled
07:05:29 <yoctozepto> #topic Backports pending reviews
07:05:45 <yoctozepto> and there are none!
07:05:54 <jopdorp> Ha!
07:05:59 <yoctozepto> #topic Release planning https://etherpad.opendev.org/p/masakari-wallaby-vptg
07:06:05 <yoctozepto> any progress on these points?
07:06:15 <jopdorp> What about the features we should get merged?
07:07:19 <yoctozepto> right
07:07:32 <yoctozepto> I guess the enabling/disabling segments is very close
07:07:46 <yoctozepto> we just need to do it nicely
07:07:53 <yoctozepto> or as nice as we can afford to
07:08:06 <yoctozepto> others are a bit further off
07:08:25 <yoctozepto> gladly this cycle was nice for its bugfixes and I have some pending ones up my sleeve
07:08:48 <yoctozepto> but that's about it from me - could not get enough priority on it
07:09:05 <yoctozepto> well, we still have 4 weeks till freeze
07:09:23 <yoctozepto> and I will accept freeze exemptions if changes are safe enough not to break the whole thing
07:11:13 <jopdorp> I did creste add a blueprint that goes with the auto reenable hosts a while ago
07:11:21 <jopdorp> Create*
07:12:03 <yoctozepto> ah, yes, I remember
07:12:15 <yoctozepto> I think I did not get much time to digest it though
07:12:29 <yoctozepto> do you have a prototype implementation to see as well?
07:12:41 <jopdorp> No
07:12:51 <jopdorp> But I could make it
07:13:12 <jopdorp> Not sure l'll be able to get it into the release though
07:13:41 <jopdorp> Although in the end a lot can be done in 4 weeks
07:14:56 <yoctozepto> that's true
07:15:26 <yoctozepto> I have one important deadline this week so can't promise anything about reviewing things
07:15:37 <jopdorp> Ok
07:15:48 <yoctozepto> (though I sometimes relax by reviewing changes)
07:16:04 <jopdorp> Should I have a look st enabling/disabling segments?
07:16:07 <yoctozepto> anyhow, please be welcome to propose a prototype
07:16:15 <yoctozepto> at least it will be ready for the next cycle
07:16:18 <yoctozepto> I think you did
07:16:21 <jopdorp> Yeah
07:16:28 <yoctozepto> you can check it again though
07:16:30 <yoctozepto> I had some comments
07:16:36 <yoctozepto> in the client
07:17:02 <jopdorp> If su is on vacation I may be able to create a new patch
07:17:19 <jopdorp> Or is that agaonst etiquette? :P
07:18:16 <yoctozepto> if you know how to tackle the issue, then please do
07:18:20 <yoctozepto> we can always revert
07:18:29 <jopdorp> Ok
07:18:31 <yoctozepto> I think the priority is to get the feature in
07:18:42 <yoctozepto> not dwell on who worked the most on it
07:19:09 <jopdorp> Yeah
07:19:35 <yoctozepto> all right
07:19:39 <yoctozepto> #topic Open discussion
07:19:52 <yoctozepto> if there is anything else... I'm all ears
07:20:18 <jopdorp> Nothing here
07:20:33 <jopdorp> Oh
07:20:37 <jopdorp> Maybe this
07:20:56 <jopdorp> Do you have any idea of the tempest tests?
07:21:22 <jopdorp> Is it our responsibility to uodate those too?
07:21:33 <jopdorp> Update*
07:21:52 <jopdorp> Do they even exist?
07:23:49 <yoctozepto> the story is the following
07:24:02 <yoctozepto> if you look closely at our functional tests job
07:24:10 <yoctozepto> it's not the expected functional tests job
07:24:21 <yoctozepto> it actually sets up the whole environment with devstack
07:24:42 <yoctozepto> and runs "functional tests"
07:25:07 <yoctozepto> they are almost like integration tests but their coverage is a tad bit less than what one would expect from integration tests
07:25:22 <jopdorp> ok
07:25:26 <yoctozepto> so we don't have masakari in tempest
07:25:37 <yoctozepto> but this job kind of aspires to do that
07:25:49 <yoctozepto> complicated, I know
07:26:01 <yoctozepto> also, there is this job that runs in NTT
07:26:23 <yoctozepto> which I have no idea what it tests because I have not received any e-mail back
07:26:33 <jopdorp> Right
07:26:43 <jopdorp> We don't have access to that source code?
07:26:53 <yoctozepto> nope
07:26:58 <yoctozepto> it's some black box
07:27:01 <jopdorp> Weird
07:27:31 <yoctozepto> yup, but NTT really is not obliged to tell us
07:27:39 <yoctozepto> I hoped they would tell me
07:27:54 <jopdorp> The functional tests are supposed to be unit tests?
07:28:03 <yoctozepto> but I am not planning to write to all NTT folks with questions, already wrote to 3 addresses
07:28:08 <yoctozepto> apart from the mailing list itself
07:28:22 <yoctozepto> nope, the unit tests are unit tests
07:28:24 <yoctozepto> they are there
07:28:33 <yoctozepto> in these tox-py36/8 jobs
07:28:50 <jopdorp> Should we at some point migrate the integration tests to tempest and write unit tests in our functional tests?
07:29:00 <yoctozepto> our current functional is between what normally is functional and integration
07:29:32 <yoctozepto> because in functional one expects to still try to isolate parts
07:29:38 <jopdorp> I guess I'm not clear on what should be the difference between functional and integration tests then
07:30:24 <yoctozepto> well, in unit tests you fake the code state (functions environments) and test code parts (functions)
07:30:47 <yoctozepto> in functional testing you fake the app environment and test the app
07:30:49 <jopdorp> Then unit tests one class, functional tests multiple classes, but all inside out project and integration tests also use a live keystone and nova?
07:31:03 <jopdorp> Our* project
07:31:06 <yoctozepto> so, the assumption would be you don't actually spin up nova for masakari functional testing
07:31:15 <jopdorp> Yeah got it
07:31:18 <jopdorp> And we do
07:31:26 <yoctozepto> functional tests actually "run the apps"
07:31:32 <yoctozepto> yup, that's it
07:31:39 <yoctozepto> and that should be done via tempest jobs
07:31:50 <yoctozepto> but then again the actual coverage of integration
07:31:55 <yoctozepto> is not that great
07:32:03 <yoctozepto> it's a good point
07:32:13 <yoctozepto> but I'm afraid we are extremely understaffed
07:32:25 <yoctozepto> to do that AND do something visible to the users
07:32:41 <yoctozepto> but still, I am tempted to improve testing coverage
07:32:46 <yoctozepto> always :-)
07:32:53 <jopdorp> Hehe
07:33:02 <jopdorp> Yeah tests can be relaxing to write
07:33:05 <yoctozepto> https://assets.mugglenet.com/wp-content/uploads/2016/01/Always.gif
07:33:16 <jopdorp> Haha
07:33:45 <yoctozepto> not only that, but the peace of mind one can get from knowing the testing pipeline and coverage are sane
07:33:52 <jopdorp> Are there coverage reports somewhere?
07:34:20 <yoctozepto> not sure, but they are only ever calculated for the unit tests I think
07:34:28 <yoctozepto> try `tox -e cover`
07:34:33 <jopdorp> Ok
07:34:54 <jopdorp> I don't understand why
07:35:34 <jopdorp> Any test framework usually has coverage reports right?
07:36:19 <jopdorp> Most stuff I do for myself has all three kinds of tests using the same framework
07:36:46 <jopdorp> Then you can generate the coverage reports separately, but also the combined coverage
07:37:02 <yoctozepto> I am pretty sure the limitation is fictitious and boils down to the general lack of order :-)
07:37:18 <jopdorp> Yeah
07:37:57 <jopdorp> Lol
07:38:09 <jopdorp> Off-topic
07:38:28 <jopdorp> I'm hapoy that they finally merged in my s3 backup driver for cinder :p
07:38:49 <yoctozepto> well, congrats!
07:39:00 <yoctozepto> then more time for masakari! :D
07:39:08 <jopdorp> Haha yes!
07:39:44 <jopdorp> I'm also happy we don't have as complicated politics as there
07:40:42 <yoctozepto> well, it would be impractical to apply anything now with the current developer base
07:40:56 <yoctozepto> so running off common sense only
07:41:20 <jopdorp> Yeah
07:42:28 <yoctozepto> thank you for the meeting jopdorp
07:42:31 <jopdorp> I think we can close the meeting
07:42:37 <yoctozepto> #endmeeting