Wednesday, 2022-06-15

*** tosky_ is now known as tosky09:42
*** dviroel|afk is now known as dviroel11:28
*** tosky_ is now known as tosky11:55
*** dasm|off is now known as dasm13:08
*** tosky__ is now known as tosky13:17
*** tosky is now known as Guest216913:29
*** tosky__ is now known as tosky13:29
*** tosky__ is now known as tosky13:58
whoami-rajat#startmeeting cinder14:00
opendevmeetMeeting started Wed Jun 15 14:00:07 2022 UTC and is due to finish in 60 minutes.  The chair is whoami-rajat. Information about MeetBot at http://wiki.debian.org/MeetBot.14:00
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.14:00
opendevmeetThe meeting name has been set to 'cinder'14:00
whoami-rajat#topic roll call14:00
fabiooliveirahi o/14:00
felipe_rodrigueso;14:00
felipe_rodrigueso/14:00
Roamer`o/14:01
jungleboyjo/14:01
simondodsleyhi14:01
rosmaitao/14:01
aneeeshp1hi14:01
caiquemello[m]o/14:01
nahimsouza[m]o/14:02
whoami-rajat#link https://etherpad.openstack.org/p/cinder-zed-meetings14:02
luizsantos[m]o/14:02
enriquetasohi14:02
whoami-rajatlooks like a good turnout14:04
whoami-rajatlet's get started14:04
toskyhi14:04
whoami-rajat#topic announcements14:04
whoami-rajatSRBAC Berlin discussion14:04
whoami-rajatso I'm not sure about the date or times of the sessions but i believe one related to service role was conducted on thursday and related to operator feedback was conducted on friday14:05
yuvalhey14:05
whoami-rajatanyway, we had a ops meetup in berlin which had a topic to discuss about our issues with the current SRBAC strategy14:05
whoami-rajatmajorly being the scopes since it has changed quite a few times in design and still the current one is not satisfactory14:06
whoami-rajatgmann, has described this in a ML thread14:07
whoami-rajat#link http://lists.openstack.org/pipermail/openstack-discuss/2022-June/028878.html14:07
whoami-rajatthe feedback etherpad for SRBAC in ops meetup is here14:07
whoami-rajat#link https://etherpad.opendev.org/p/rbac-operator-feedback14:07
whoami-rajathonesty, i wasn't able to derive any conclusion or major points to highlight here14:07
whoami-rajatso feel free to take a look at the etherpad discussion section14:08
whoami-rajatanother session related to service role was held on thursday last week14:08
whoami-rajat#link https://etherpad.opendev.org/p/deprivilization-of-service-accounts14:08
whoami-rajatagain, the etherpad says very less and until a recording is available, I couldn't derive much concrete items from the etherpad14:09
rosmaitame neither14:09
whoami-rajati was hoping rosmaita would but hard luck :/14:09
whoami-rajatlet's see if TC posts an update about it14:10
rosmaitathe policy pop-up team did not meet yesterday, so hoping to get some info on thursday at the TC meeting14:10
gmannyeah, I am consolidating about all the feedback from various place and ops meetup/forums etc14:10
whoami-rajatgmann, great thanks14:11
gmannand also will send the meeting schedule soon to decide the next step14:11
whoami-rajatsounds like a plan so let's wait for further discussions14:11
whoami-rajatthanks gmann for an update14:12
rosmaitagmann: my impression is that ironic absolutely wants scope, but maybe it can be optional for other projects?14:12
rosmaita(if it's too complicated to answer, we can discuss at the tc meeting)14:12
gmannrosmaita: may be but ironic can be exception but any other inter-dependent projects should not have scope as individual. but let's discuss in policy popup meeting14:12
gmannyeah14:12
whoami-rajatthanks again, let's move on to our next announcement14:13
whoami-rajatnext, cinderlib for yoga must be released by 23 June14:13
whoami-rajatrosmaita, that's you14:13
rosmaitajust wanted to remind everyone, i will talk about some issues later14:14
whoami-rajatack, I received one reminder from herve about the cinderlib release14:14
whoami-rajatthey've proposed a patch14:14
whoami-rajat#link https://review.opendev.org/c/openstack/releases/+/84570114:14
rosmaitayeah, please put a -1 on that14:15
whoami-rajatand I've told him that after discussing all the issues we've in cinderlib, me or rosmaita will add a comment to the patch14:15
rosmaitaalso, there's https://review.opendev.org/c/openstack/releases/+/84210514:15
whoami-rajatah, this one is old ...14:16
whoami-rajatI will ask him to abandon the new one in favor of this one then14:16
rosmaitasounds good, just -1 both so that there's no confusion14:17
whoami-rajatyep, will do, thanks14:17
whoami-rajatnext annoucement, spec freeze is 24 June14:17
whoami-rajatso i see the quota spec by geguileo still needs an update14:18
rosmaitahas anyone looked at the task status field proposal?14:18
simondodsleyalso this one needs to be addressed: https://review.opendev.org/c/openstack/cinder-specs/+/81855114:18
simondodsleythat's the one14:19
rosmaitathat's the one14:19
rosmaitajinx!!!14:19
whoami-rajatalso the SRBAC one needs to be updated after we've all the discussion points from ops meetup (maybe will get a Spec feeze exception)14:19
simondodsleyWalt has an issue with it14:19
simondodsleyhemna are you here?14:19
geguileowhoami-rajat: yeah, sorry I was busy with the nvmeof and backup memory usage stuff14:20
* fabiooliveira oh no, you're jinxed -- kidding14:20
geguileowhoami-rajat: those are ready now, so I should be able to go back to it14:20
simondodsleythe task status one really needs to be either approved or not so they can start the coding. This has been hanging around since Yoga14:21
whoami-rajatgeguileo, np, just wanted to know we're on track as next week is spec freeze14:21
rosmaitanext time someone sees hemna in #openstack-cinder, please ask him to go back to https://review.opendev.org/c/openstack/cinder-specs/+/818551 and respond to their responses14:21
whoami-rajatgeguileo, great14:21
geguileowhoami-rajat: who needs sleep!!14:21
whoami-rajat:D sorry about all the work items you've got in this cycle ... and everything is IMPORTANT!!14:22
geguileolol14:22
simondodsleythere are a lot of old specs out there - we need to either kill them or retarget to zed14:23
whoami-rajatas rosmaita said, please followup with hemna on the spec and i will also take a look by this week14:23
rosmaitaspeaking of the memory usage stuff, geguileo has an interesting patch up: https://review.opendev.org/c/openstack/devstack/+/84580514:23
whoami-rajatthe numbers look fascinating, 1GB -> 130 MB14:24
rosmaitaplus, it's not cinder-backup's fault!14:24
rosmaita(that's the best part)14:24
enriquetaso++14:24
geguileolol14:24
geguileoyeah, not our fault for once14:24
whoami-rajatyeah, another not a cinder issue14:25
jungleboyj\o/14:25
whoami-rajatsimondodsley, good idea, i will take a look at specs that are not relevant for Zed and ask them to be retargeted or abandoned14:26
whoami-rajatok so let's move on to topics14:26
simondodsleyi've already asked for retargets but there have been no responses14:26
whoami-rajatoh, then we probably should abandon them after a certain amount of time but not too sure about it14:27
whoami-rajatrosmaita, what do you think ? ^14:27
rosmaitawhoami-rajat: yes, we should abandon anything maybe older than 2 cycles14:28
jungleboyjrosmaita: ++14:28
rosmaitawith a note, "feel free to restore if you want to keep working on this"14:28
whoami-rajatwe've got 2 PTLs approval on this so let's move on with this strategy ^14:28
whoami-rajatok moving on to topics14:29
whoami-rajat#topic Reviews request14:29
whoami-rajatenriquetaso, that's you14:29
enriquetasoHey14:29
whoami-rajat#link https://review.opendev.org/c/openstack/devstack-plugin-ceph/+/782624/14:29
enriquetasoJust asking for reviews again :P 14:29
enriquetaso I really need some reviews in this to continue working on more tempest patches.14:29
enriquetasoI'd like to see the mimic client on the CI before submitting more tempest test14:30
rosmaitareminder: all cinder-core are also devstack-plugin-ceph cores14:30
enriquetasothanks! That's all for me!14:30
enriquetasoI think we discussed this patch a few time ago on this meeting14:30
enriquetasoand could be ready to merge 14:30
whoami-rajatthe comment i read in the patch might not be too accurate14:31
whoami-rajat# Enables new features such as Clone v2 API, which allows proper handling of14:31
whoami-rajat# deleting snapshots with child clone images.14:31
whoami-rajatbut will add a comment to the patch14:31
whoami-rajatso cores, please take a look as it's blocking work ^14:31
enriquetasowhoami-rajat++14:31
enriquetasothanks, I'll update the commit msg after your review 14:32
whoami-rajatack thanks14:32
whoami-rajatmoving on14:32
whoami-rajat#topic Four issues with cinderlib14:32
whoami-rajatrosmaita, that's you14:32
rosmaitalot to talk about here, but mostly informative (i believe everything is close to a solution if we agree)14:32
rosmaitaso, the previous PTL seems to have left cinderlib in a heck of a state14:32
rosmaitaissue #1: cinderlib CI14:32
rosmaitacinderlib is a cycle-trailing release, so master is still yoga development14:32
rosmaitahasn't been a problem in the past, but zed doesn't support py36 anymore, while yoga does14:33
rosmaitaso we started hitting gate failures due to testing cinderlib (yoga development) with master (zed development) upper-constraints14:33
rosmaitafixed by including overrides in .zuul.yaml14:33
rosmaitapart one (merged): https://review.opendev.org/c/openstack/cinderlib/+/84517014:33
rosmaitapart two: https://review.opendev.org/c/openstack/cinderlib/+/84527214:33
rosmaita(job wasn't failing, but I think that was luck, and we should be consistent about these changes)14:33
rosmaitaso, just need reviews on part two ^^14:34
rosmaitaand the issue will be solved!14:34
rosmaitaissue #2: unconstrained builds14:34
rosmaitatox.ini is set up so that we install cinder and os-brick from source14:34
rosmaitaos-brick is in upper-constraints, so if we use upper-constraints, we can't install it14:34
rosmaita(because the development version exceeds what's in upper-constraints)14:34
rosmaitaat the same time, if we don't constrain cinderlib for testing, we really don't know what library versions are actually being used14:34
rosmaitaso we really do want to constrain it14:34
rosmaitaproposed solution is https://review.opendev.org/c/openstack/cinderlib/+/84560714:34
rosmaita(which needs reviews)14:35
rosmaitait creates a local-upper-constraints.txt that doesn't include os-brick14:35
rosmaitait gets generated as part of the tox install_command14:35
rosmaitayou can override the file that's used with the CINDERLIB_CONSTRAINTS_FILE environment var14:35
rosmaitaI decided not to hide local-upper-constraints.txt in the tox temp dir so you can see exactly what's being used14:35
rosmaitathat filename is added to .gitignore, so it shouldn't bother you at all14:35
rosmaitathe reason why we're not using the standard TOX_CONSTRAINTS_FILE environment var is to make this change work with zuul14:35
rosmaitazuul always overrides TOX_CONSTRAINTS_FILE to use upper-constraints directly from its install of openstack/requirements14:35
rosmaita(this makes it possible to use Depends-on when testing upper-constraints changes)14:35
rosmaitaand it overrides it aggressively, can't change this in our .zuul.yaml14:36
rosmaitaso we have to use CINDERLIB_CONSTRAINTS_FILE14:36
rosmaitathe downside is that we can't use Depends-on to test upper-constraints changes in the gate14:36
rosmaitabut i don't think this is a big deal because cinderlib is a cycle-trailing release14:36
rosmaitaso any problems will most likely be caught earlier by cinder14:36
rosmaitaand you can always test locally by downloading the patched u-c file and setting CINDERLIB_CONSTRAINTS_FILE14:36
rosmaita(though you have to remember to remove os-brick from the patched file)14:36
rosmaitaso that's how the patch works, please review and leave questions etc:14:36
rosmaitahttps://review.opendev.org/c/openstack/cinderlib/+/84560714:37
rosmaitathe nice thing is that that patch works for both tox and zuul14:37
rosmaitaissue #3: not running requirements-check job14:37
rosmaitathe standard requirements check template is not set up to handle trailing releases14:37
rosmaitabut we only have 3 requirements:14:38
rosmaitahttps://opendev.org/openstack/cinderlib/src/branch/master/requirements.txt14:38
rosmaitaso, this file ^^ is interesting because it contains cinder14:38
rosmaitait's needed for when someone installs cinderlib from pypi14:38
rosmaitawe don't actually use the requirements.txt in the cinderlib tox.ini14:38
rosmaitabecause we install cinder and os-brick from source14:38
rosmaitaand importlib-metadata is used by cinder, so we get it that way14:38
rosmaitaso my proposal is that the PTL just check manually to make sure that requirements.txt is correct14:38
rosmaitaos-brick and importlib-metadata are in global-requirements14:39
rosmaitabut ... cinder is not14:39
rosmaitaand it's not allowed in there14:39
rosmaitafor info about this if you care, see this discussion in #openstack-requirements:14:39
rosmaitahttps://meetings.opendev.org/irclogs/%23openstack-requirements/%23openstack-requirements.2022-06-08.log.html#t2022-06-08T15:36:5514:39
rosmaitaif cinderlib starts using more requirements, we can revisit this14:39
rosmaitabut for now, I propose we do nothing14:39
rosmaita(just wanted to make sure everyone understands what's going on)14:39
rosmaitaok, finally14:40
rosmaitaissue #4: not using released versions of cinder, os-brick in CI14:40
rosmaitawe probably had this discussion when cinderlib CI was first set up for the train release, but I don't remember the reasons14:40
rosmaitaso the issue is that all our CI is using cinder and os-brick source, so possibly using unreleased changes14:40
rosmaitawe could add more jobs14:40
rosmaitaor, we could just make sure that when we release yoga cinderlib14:40
rosmaitawe also release yoga cinder and os-brick (if they contain any unreleased changes)14:40
rosmaitathen we'll know that cinderlib has not been relying on any unreleased code to pass its CI14:41
rosmaitathat's it ... so to summarize14:41
rosmaitaissue #1 pretty much solved14:41
rosmaitaissue #2 probably solved?14:41
rosmaitaissues #3, #4 ... i propose we do nothing14:42
rosmaitasorry that was a lot of text to dump in here14:42
rosmaitaany questions?14:42
tosky(I voted -1 on the second patch of #1 but it's either easily solvable or I'm plainly wrong)14:42
geguileorosmaita: regarding the cinderlib requirements, it should *never* have any more than what we currently have14:42
geguileoso it should be ok leaving it as it is (like you propose)14:42
rosmaitaworks for me!14:42
whoami-rajatfor #3, i think it's OK to manually check if right versions of cinder and os-brick are mentioned in requirements.txt unless someone thinks otherwise14:42
rosmaitayeah, the alternative is to hack the requirements file like i did with upper-constraints ... don't think it's worth it, though14:43
geguileorosmaita: I don't see the issue for #4 14:43
whoami-rajatfor 3 requirements? don't think so and doesn't add much burden on me as well so no problem at all14:44
geguileorosmaita: when both are working on the same release it makes sense to run it against master, since we want them to keep in sync and not find surprises14:44
geguileorosmaita: once os-brick releases we pin it to in cinderlib tox.ini to the stable branch, and the same thing when cinder releases14:45
geguileoand then once cinderlib releases we can unpin those 214:45
rosmaitaright, it's just that it could be possible that some stuff has been merged to cinder or os-brick stable/yoga that we are testing with14:45
geguileoI believe that's mostly what we've been doing14:45
rosmaitaand if someone installs cinderlib from pypi, they get released versions of cinder, os-brick 14:46
geguileotrue14:46
rosmaitait's pretty unlikely14:46
geguileoI don't anticipate many issues there though14:46
rosmaitabut to be safe, we can just release new cinder and os-brick at the same time14:46
rosmaitayeah, i am just being over-cautious14:47
geguileoyeah, I just feel bad giving extra work with those 2 additional releases14:47
whoami-rajatrosmaita, do you mean releasing stable/yoga of cinder and os-brick?14:47
rosmaitareleases are cheap ... testing is hard!14:47
rosmaitawhoami-rajat: yes, exactly14:47
jungleboyj:-)14:47
rosmaitaby the way, i meant to say something about cinderlib for people here who are unfamiliar with it14:48
whoami-rajatwe can do it but if we think we're good with what we currently have then maybe not required14:48
whoami-rajatok that doesn't seem like a big issue to discuss right now, i think the focus should be more on #1 and #214:49
rosmaitacinderlib is used by Ember-CSI which is a container storage interface for kubernetes14:49
whoami-rajatand thanks rosmaita for finding out the issues and providing a verbose summary14:49
rosmaitaso you can use the cinder drivers without having to run cinder as a service14:50
rosmaitayeah, verbosity is my middle name14:50
rosmaitathat's all from me14:50
whoami-rajatlol14:50
whoami-rajatgreat, so i guess that's all we had for topics14:50
whoami-rajatlet's move to open discussion14:50
simondodsleyAnyone else seeing tempest test `tempest.api.compute.admin.test_volume_swap.TestMultiAttachVolumeSwap` failing with `ValueError: Multiple pingable or sshable servers not supported at this stage`? Seems to be ever since https://review.opendev.org/c/openstack/tempest/+/842921 merged14:51
whoami-rajat#topic open discussion14:51
aneeeshp1Hi, I am representing Fungible (https://www.fungible.com/product/nvme-over-tcp-fungible-storage-cluster/) and attending this meeting for the first time.14:51
aneeeshp1Just wanted to talk about a new driver for Fungible storage backend.14:51
aneeeshp1I have submitted a blueprint for the new driver (https://blueprints.launchpad.net/cinder/+spec/fungible-volume-driver)14:52
aneeeshp1CI setup in progress. Expected to be ready by end of this month. Planning to submit a patch once the CI is ready.14:52
aneeeshp1Can this be targeted for Zed?14:52
rosmaitaaneeeshp1: welcome!14:52
whoami-rajataneeeshp1, Welcome!14:52
geguileoaneeeshp1: welcome to the cinder meetings!14:52
aneeeshp1Thank you!14:52
jungleboyjWelcome.  :-)14:52
fabiooliveirawelcome \o/14:52
whoami-rajataneeeshp1, sine you've filed a blueprint, you're already on the right track, one question, do you have a patch up for the new driver?14:53
rosmaitaaneeeshp1:  https://releases.openstack.org/zed/schedule.html#cinder-new-driver-merge-deadline14:53
whoami-rajatah you already said it will be pushed once CI is ready, my bad14:53
aneeeshp1whoami-rajat: not yet. Will be ready by end of this month.14:53
whoami-rajataneeeshp1, so currently our deadline for driver merging is 15th July, maybe enough time to review the change but will be good if you can try to get it done earlier14:54
whoami-rajat#link https://releases.openstack.org/zed/schedule.html#z-cinder-driver-deadline14:54
rosmaitahow many drivers are proposed at this point? i am losing count14:55
whoami-rajatah, rosmaita shared this already, I'm skipping some messages ...14:55
whoami-rajatmore than we can review?14:55
geguileoaneeeshp1: you can push the patch before upstream CI is ready14:55
whoami-rajat6-7 probably14:55
aneeeshp1Can the patch review start before the CI is ready. I might be able to submit the patch earlier, but CI will take some time (end of the month)14:55
geguileoyes it can14:55
aneeeshp1Thank you geguileo. I will do that14:55
whoami-rajatyes, it won't be merged unless the CI is reporting but that doesn't block reviewing the driver patch14:56
aneeeshp1Okay thanks14:56
aneeeshp1I will create patch ASAP.14:56
whoami-rajatgreat, thanks for your contribution14:56
aneeeshp1thank you all14:56
rosmaitalooks like 8 new drivers14:57
whoami-rajatmake sure to add it to the work items section in your blueprint14:57
aneeeshp1sure14:57
whoami-rajatrosmaita, wow, maybe the highest I've seen in a cycle14:57
jungleboyjYes.14:57
jungleboyjSince the old days when we were the hot new thing in town.14:58
whoami-rajatI will create etherpad for spec and drivers to prioritize them14:58
enriquetasocool14:58
Roamer`hi, so real quick (I hope... unless there are any objections and maybe I should have put this in the schedule)... so you may remember that in the May 25h video meeting I brought up a problem with the StorPool driver keeping Glance images in a different Cinder pool (underlying StorPool template) than the volumes the users wish to create, and there seemed to be some consensus that instead of 14:58
Roamer`every driver reimplementing half of the workflow for creating an image out of a volume, it might be easier to add a driver capability "I know how to clone volumes efficiently even into a different pool"... so today I filed https://blueprints.launchpad.net/cinder/+spec/clone-across-pools and what do people think about the name of the capability?14:58
Roamer`I have already started working on it (we have to do something like this at a customer installation and this option, a driver capability, will be *much* cleaner than what we have now), I guess I will have something ready for review in a day or two14:59
whoami-rajatalso one announcement i forgot ... we've the festival of XS reviews this Friday, but i will be sending a reminder to ML anyway14:59
enriquetasooh,i wont attend this XS review festival because i'm on AR Holiday :(15:00
rosmaitawow, third friday of the month has arrived really fast!15:00
jungleboyjrosmaita:  Yeah, how did that happen already?  :-)15:01
whoami-rajatRoamer`, thanks for providing the update15:01
whoami-rajatenriquetaso, ah shoot, but no problem15:01
whoami-rajatrosmaita, yeah really, i thought it was the second one this Friday15:01
whoami-rajatwe've passed the time limit for the meeting15:01
whoami-rajatso let's wrap it up15:02
whoami-rajatthanks everyone15:02
whoami-rajat#endmeeting15:02
opendevmeetMeeting ended Wed Jun 15 15:02:09 2022 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)15:02
opendevmeetMinutes:        https://meetings.opendev.org/meetings/cinder/2022/cinder.2022-06-15-14.00.html15:02
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/cinder/2022/cinder.2022-06-15-14.00.txt15:02
opendevmeetLog:            https://meetings.opendev.org/meetings/cinder/2022/cinder.2022-06-15-14.00.log.html15:02
jungleboyjThanks!15:02
*** dviroel is now known as dviroel|lunch15:18
*** dviroel|lunch is now known as dviroel16:15
*** dviroel is now known as dviroel|afk18:47
*** dasm is now known as dasm|off21:09

Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!