Wednesday, 2021-12-15

*** hemna0 is now known as hemna07:37
*** dviroel_ is now known as dviroel|rover11:14
*** kopecmartin_ is now known as kopecmartin13:41
*** whoami-rajat__ is now known as whoami-rajat14:00
rosmaita#startmeeting cinder14:02
opendevmeetMeeting started Wed Dec 15 14:02:13 2021 UTC and is due to finish in 60 minutes.  The chair is rosmaita. Information about MeetBot at http://wiki.debian.org/MeetBot.14:02
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.14:02
opendevmeetThe meeting name has been set to 'cinder'14:02
jungleboyjo/14:02
rosmaita#topic roll call14:02
shoffmannhi14:02
fabiooliveirahi o/14:02
toskyhi14:02
eharneyhi14:02
whoami-rajatHi14:02
nahimsouza[m]o/14:02
rosmaitaapologies for the late start, i was in #openstack-meeting14:02
walshh_hi14:02
SachinMorehi14:02
e0nehi14:03
TusharTgitehi14:03
felipe_rodrigueshi14:04
rosmaitaok, let's get started14:04
rosmaita#topic announcements14:04
rosmaitafirst, about our meetings over the next few weeks14:05
whoami-rajat#link https://etherpad.opendev.org/p/cinder-yoga-meetings14:05
rosmaitathanks!14:05
rosmaitai am very unorganized this moringin14:05
enriquetasohi14:05
rosmaitaok, 29 December will be no meeting14:05
rosmaitaso that means next week is our last of the month14:05
rosmaitaquestion: should we do it in video?14:06
enriquetasowe can toast the new year :P14:06
rosmaitai'm inclined to, but in case anyone actually hates video meetings, we don't have to14:06
jungleboyjHow many people will be around next week?14:06
rosmaitajungleboyj: that is a good question, too14:06
jungleboyj:-)14:06
rosmaitai will be around14:06
eharneyi'm out next week14:06
jungleboyjI will be out.14:07
fabiooliveirai will be out also14:07
rosmaitaok, maybe we should take enriquetaso's idea and have quick meeting/social hour for anyone around14:07
rosmaitaunless that is a stupid idea14:08
rosmaitameaning my idea, not enriquetaso's14:08
rosmaitai am really not at my best this morning14:08
enriquetasohaha14:08
rosmaitaok, maybe let's do this ... i will cancel meetings 22 and 29 December, but will be around in irc at that time in case people have questions, want to discuss anything14:09
rosmaitaok, the first wednesday in 2022 is 5 January14:10
rosmaitai will not be here14:10
rosmaitaso, i need a volunteer to run the meeting, or we can cancel that one too14:10
whoami-rajatI can run if people will be around14:10
rosmaitathanks whoami-rajat14:10
jungleboyjI will be back then. 14:10
rosmaitaeharney: enriquetaso: will you be around?14:11
enriquetasoyes14:11
eharneyyes14:11
rosmaitaok, great, we will have that meeting14:11
rosmaitanext announcement14:11
rosmaitaFestival of XS Reviews on Friday14:12
rosmaita1400-1600 utc14:12
rosmaitain meetpad, connection info here: http://eavesdrop.openstack.org/#Cinder_Festival_of_XS_Reviews14:12
rosmaitaand, finally, cinderlib 4.1.0 was released on Monday14:12
rosmaitaso now cinderlib master is open for Yoga development14:13
rosmaita#topic yoga testing runtime update14:13
whoami-rajat\o/14:13
rosmaitathere was a really long ML thread about this that spanned November and December14:14
rosmaitasummary is here:14:14
rosmaita#link http://lists.openstack.org/pipermail/openstack-discuss/2021-December/026164.html14:14
rosmaitathe short story is that openstack will support python 3.6, 3.7, 3.8, 3.914:14
rosmaitathe testing template has been updated to test 3.6 and 3.914:14
rosmaitaand passing those 2 will be taken as an indicator that 3.7 and 3.8 would pass too14:15
rosmaitathe testing template only covers unit tests14:15
rosmaitaso we have to patch .zuul.yaml to get the correct functional tests14:16
rosmaitaTushar has put together a bunch of patches14:16
rosmaitathe cinderclient functional-py39 is passing running on centos-9-stream14:16
rosmaitabut the cinderclient-brick-ext functional-py39 is failing on a pip constraint satisfaction problem14:17
rosmaitai think we need to get advice from #opendev team about what nodes we should use for py3914:17
rosmaitawe had to use centos-8-stream to run the py36 functional tests14:18
rosmaitabut i am not sure what the consensus is for 3.914:18
rosmaita(and the platform may actually have nothing to do with that failure)14:18
*** AndreBeltrami[m] is now known as andrebeltrami[m]14:19
rosmaitaalso, the ML thread mentioned py 3.10 job in the template as non-voting14:19
rosmaitabut i don't think that has been implemented yet14:19
rosmaitai will follow up with gmann and see what the plan is14:19
rosmaitaotherwise, i'll put up our own patches for non-voting py 3.10 so we can get early warning if we don something incompatible14:20
rosmaita#topic specs review14:21
rosmaita#link https://etherpad.opendev.org/p/cinder-yoga-specs14:21
rosmaitaok, let's start with the easiest one14:21
rosmaitahttps://review.opendev.org/c/openstack/cinder-specs/+/81867114:21
rosmaitathat moves a previously accepted spec to Yoga14:22
rosmaitalooks like no one has expressed any concerns, so i think that is ready to go14:22
rosmaitathis next one isn't easy, but it's got two +2s14:23
rosmaitahttps://review.opendev.org/c/openstack/cinder-specs/+/81969314:23
rosmaitaGorka put a -W on it because he wanted to get performance data on the change14:23
rosmaitaso i propose we give him a spec freeze exception while he decides whether the numbers indicate that the proposal is OK, or will need a revision14:24
whoami-rajat+1, that is a great improvement over what our quota system currently has14:25
rosmaitayes, hopefully the performance numbers won't be any worse than currently14:26
rosmaitathere's an analysis in the spec about how it reduces the number of transactions14:26
rosmaitawhich is good, but maybe the counting is more problematic than we thought14:27
* jungleboyj swears that I read that spec but don't see my vote.14:27
rosmaitabtw, the recording of Gorka explaining the spec is available14:28
rosmaita#link https://youtu.be/lrg9T8l8g4g14:28
jungleboyjBased on the discussion I think it is a good thing.14:28
rosmaitayeah, i hope the questionable performance was some kind of anomaly14:29
rosmaitabut it was enough to make Gorka hold up the spec14:29
rosmaitathough he has very high standards, so maybe it isn't that big a deal14:29
rosmaitawe'll see14:29
rosmaitaok, looking at specs with no -1s14:30
rosmaitaSupport revert any snapshot to the volume14:30
rosmaita#link https://review.opendev.org/c/openstack/cinder-specs/+/73611114:30
rosmaitai have a question about how we will microversion this, though that could be addressed in a followup14:31
rosmaitaanyway, please react on the gerrit review14:31
rosmaitaUpdate original resource's az14:32
rosmaita#link https://review.opendev.org/c/openstack/cinder-specs/+/77843714:32
rosmaitai haven't looked at the update yet14:32
rosmaitawe agreed that this is a good idea in principle, but i am not sure all the details have been worked out sufficiently14:33
rosmaitaAdd volume reimage spec14:33
rosmaita#link https://review.opendev.org/c/openstack/cinder-specs/+/80962214:34
rosmaitathat's whoami-rajat's spec, so he can't review it himself14:35
whoami-rajat:D14:35
whoami-rajattrue14:35
jungleboyj:-)14:36
rosmaitai had a question on that one about the name of a flag, but otherwise it looks fine and the nova team is OK with it14:36
rosmaitaactually, they were quite supportive at the PTG14:36
whoami-rajatyeah, i think they really need this feature14:37
rosmaitai just realized that i should be clear about this discussion (about all the specs)14:37
rosmaitaif anyone has a serious reservation, please speak up!14:37
rosmaitaOs-brick NVMeoF connector multipathing spec14:38
rosmaita#link https://review.opendev.org/c/openstack/cinder-specs/+/81561314:38
rosmaitathis looks good, I think most of the questions have been answered14:39
rosmaitathere was one point Gorka made that Zohar didn't agree with, so look for that14:40
rosmaitaotherwise, it's ready14:40
jungleboyjWhen is Spec Freeze again?14:41
rosmaitaFriday14:41
rosmaitaAdd spec to optimize upload volume for RBD backend14:41
rosmaitahttps://review.opendev.org/c/openstack/cinder-specs/+/81036314:42
jungleboyjOk.14:42
rosmaitai have a question about whether this should be controlled from the glance side or the cinder side14:42
rosmaitabut maybe that's just me ... need some other people to look this over14:42
rosmaitawe discussed at the PTG and everyone was generally supportive14:42
eharneyi'm curious if that specs makes assumptions about permissions to access the glance pool etc14:43
rosmaitaeharney: that is a good question14:43
eharneyi think it might need more work to handle a separate set of credentials/keys etc, will take a look14:44
whoami-rajateharney, that needs to access glance pool as it creates the clone in glance pool14:44
rosmaitaImplement backend QOS for rbd driver14:45
rosmaitahttps://review.opendev.org/c/openstack/cinder-specs/+/82003014:45
rosmaitai haven't looked at the revised version, the open issue was a question about mirroring14:46
rosmaitabut i think the general idea is fine14:46
rosmaitaso, needs a look from anyone interested in Ceph ^^14:46
eharneyi wonder how this intersects with the _per_gb qos fields14:47
rosmaitai think all the remaining specs have unanswered -1s14:47
rosmaitaso cinder-core team, please review the specs on the etherpad14:48
rosmaita#link https://etherpad.opendev.org/p/cinder-yoga-specs14:49
rosmaitabut we should keep an eye out for revisions, since the deadline is not until friday14:49
rosmaitais there anyone here who would like to speak about a spec they have proposed?14:49
rosmaitaok, moving on14:51
rosmaita#topic mypy patch of the week14:51
rosmaitawell, i am happy to say that our previous patch of the week merged!14:52
rosmaitabut, it was the R-20 patch of the week14:52
rosmaitaand this is week R-1514:52
enriquetaso\o/14:52
rosmaitaeharney: do you have a particular patch in mind for this week?14:52
eharneywell, https://review.opendev.org/c/openstack/cinder/+/787195 had two +2s and a +W back in April, so maybe an easier one14:53
rosmaitaeasy is good!14:53
rosmaitaok, let's adopt that one14:53
rosmaitalooks like it was jungleboyj and me14:54
*** lamt_ is now known as lamt14:58
*** simondodsley_ is now known as simondodsley14:58
*** angiolucci_ is now known as angiolucci14:59
* whoami-rajat thinking if i lost connection or meeting is frozen15:00
* enriquetaso is waiting too15:01
* fabiooliveira guess is frozen 👀15:01
*** walshh__ is now known as walshh_15:01
*** elenalindq_ is now known as elenalindq15:01
angiolucciI see some people were disconnected, some still rehoining15:02
enriquetasosomething happened 15:02
fabiooliveiramay be some issue with irc15:02
eharney#endmeeting15:05
opendevmeetMeeting ended Wed Dec 15 15:05:59 2021 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)15:06
opendevmeetMinutes:        https://meetings.opendev.org/meetings/cinder/2021/cinder.2021-12-15-14.02.html15:06
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/cinder/2021/cinder.2021-12-15-14.02.txt15:06
opendevmeetLog:            https://meetings.opendev.org/meetings/cinder/2021/cinder.2021-12-15-14.02.log.html15:06
*** dviroel|rover is now known as dviroel|rover|lunch15:47
*** dviroel|rover|lunch is now known as dviroel|rover16:32
*** dviroel|rover is now known as dviroel|out21:11

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