Wednesday, 2024-02-07

whoami-rajat#startmeeting cinder14:00
opendevmeetMeeting started Wed Feb  7 14:00:05 2024 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
happystackerHello !14:00
rosmaitao/14:00
sp-bmilanovo/14:00
Luzio/14:00
crohmanno/14:00
akawaio/14:01
simondodsleyo/14:01
msaravanHi14:01
whoami-rajat#link https://etherpad.opendev.org/p/cinder-caracal-meetings14:02
eharneyo/14:03
zaitcevo/14:03
whoami-rajatwelcome everyone14:04
whoami-rajatlet's get started with announcements14:04
whoami-rajat#topic announcements14:04
whoami-rajatMidcycle - 214:04
whoami-rajat#link https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/BJ7EFMPBDNZMDVYWASC7CFY5GXW4GEGX/14:04
whoami-rajatDetails:14:04
whoami-rajatDate: 14th February, 202414:04
whoami-rajatTime: 1400-1600 UTC14:04
whoami-rajatMeeting link: https://meet.google.com/cqw-kbpn-sxx14:04
whoami-rajatEtherpad: https://etherpad.opendev.org/p/cinder-caracal-midcycles14:04
whoami-rajatWe only have a week so please add your topics14:04
whoami-rajatget ready for 2 hour meeting next week!14:05
whoami-rajatnext, Releasenote job breaking and fix14:06
whoami-rajat#link https://review.opendev.org/c/openstack/cinder/+/90783614:06
whoami-rajatAs our transition from yoga-eom, we deleted the stable/yoga branch and created unmaintained/yoga14:06
whoami-rajatThis caused a failure in our releasenotes job because we used the tag stable/yoga14:06
whoami-rajatI approved the patch but Brian left a comment that we should use yoga-eom instead of unmaintained/yoga as our releasenote tag14:07
simondodsleyLOL - I just noticed that and was putting it in as a topic...14:07
whoami-rajati also noticed today itself, but the release team was ahead of us14:07
rosmaitaMe too, just noticed this morning14:07
rosmaitaso here's why it matters14:08
whoami-rajatso unmaintained/yoga is synchronous with stable/yoga since it will be an active branch accepting changes14:08
rosmaitawe don't have responsibility for what goes on in unmaintained/yoga14:08
happystackershould we update https://releases.openstack.org/ as well?14:08
rosmaitaso i think we should end the yoga series release notes with the last stuff we have done14:08
rosmaitawhich would be everything up to the yoga-eom tag14:09
simondodsleythere are 43 other projects using unmaintained/yoga14:09
rosmaitayeah, but we Argonauts do not follow the crowd!14:09
whoami-rajathappystacker, that should be updated yes, but i guess release team will already be looking into it14:09
simondodsleywe should be in sync even if it is technically wrong14:09
happystackeroh ok good14:09
whoami-rajatnot sure if frickler is around but would be good to get his opinion on this14:10
simondodsleyrosmaita s/crowd/heard/14:10
rosmaitaabout being in sync ... there is a bit of leeway with respect to the unmaintained branches14:10
rosmaitaironic, for example, wants to actively manage theirs14:10
rosmaitaso it makes sense for them to include those changes in the official release notes14:10
rosmaitasince we don't want to, i think we should stop our release notes with our last official actions in that branch14:11
fricklerwhoami-rajat: sorry having another meeting in parallel, not sure about the question?14:11
rosmaitafrickler: has to do about whether to publish unmaintained/yoga stuff in the cinder release notes14:11
whoami-rajati agree, the whole unmaintained thing is new so not really sure about the right thing, but what rosmaita said makes sense to me, if the project wants to be active in their unmaintained branch then unmaintained tag is good else eom should be a consideration14:12
whoami-rajatfrickler, so rosmaita suggested that we should use yoga-eom instead of unmaintained/yoga as the tag since cinder team won't be actively looking after the unmaintained branch14:12
frickleryes, renos should still include unmaintained notes, at least that seemed to be the major opinion of the tc14:12
rosmaitawe did not actually consider that issue at the TC14:13
rosmaitaat least not that i'm aware of14:13
fricklerwell ok ... of tc members14:13
fricklerbut yeah, bring it up at the next meeting, certainly worth more discussion14:14
rosmaitain the meantime, i am in favor of the -eom tag, it fits more with the "opt in" nature of the unmaintained stuff14:14
whoami-rajatI'm pro the discussion but would also like the cinder gate to be fixed in the meantime :D14:14
simondodsley+114:14
whoami-rajatI don't have any objections there14:15
whoami-rajatalso we can always update it after the discussion14:15
whoami-rajatso +1 from me14:15
rosmaita+1 to which?  unmaintained or eom?14:15
whoami-rajateom14:15
rosmaita\o/14:16
rosmaitalet's do it, and we can set a good example14:16
rosmaita:D14:16
whoami-rajatyeah, I'm convinced with the idea that this should be an opt in based on if the project wants to maintain their branches14:17
whoami-rajatelse there will be questions on releasenotes about changes that we have no idea about14:17
rosmaitaok, i'll update the patch (so if there are complaints, they'll be about me)14:17
whoami-rajatthanks, i can share the complaints :D14:18
whoami-rajatokay, so we will be going with the eom tag in releasenotes instead of the unmaintained one14:19
rosmaitaok, you can go ahead and ninja it in to get the gate un-broken14:19
rosmaitaand i guess we have to backport it?  i first notice this in looking at a stable/zed patch , https://review.opendev.org/c/openstack/cinder/+/90625914:20
whoami-rajatI remember frickler  confirmed that stable branches should be automatically fixed by this14:22
whoami-rajatthere is some reno magic that i have no clue about14:22
whoami-rajatbut he did test it in 2023.2 branch and gate was passing14:22
whoami-rajatokay moving on to the next announcement since we have a bunch of topics lined up14:23
frickleryes, see https://review.opendev.org/c/openstack/designate/+/908290 , so please double check after merging the fix in master before doing too many patches14:23
whoami-rajatthanks frickler !14:23
whoami-rajatnext, Eventlet update14:23
whoami-rajat#link https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/7PJZ4WYRS2B3PIZ5KYFL6V3JWYL4V5FQ/14:23
whoami-rajat#link https://review.opendev.org/c/openstack/requirements/+/90724314:24
rosmaitaok, cool14:24
whoami-rajatThe version 0.35.0 broke docs job in manila and now 0.35.1 is released and merged in our UC14:24
whoami-rajatThe latest version of eventlet is expected to work across openstack14:24
whoami-rajatWe haven't seen any failures in cinder related to it so looks good for us14:24
whoami-rajatnext, oslo feature freeze14:25
whoami-rajat#link https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/DB7EAOKIHXCULKISSFZCYTIUMEATPGQQ/14:25
whoami-rajatdeadline: R-7: Feb 12 - Feb 1614:25
whoami-rajatIf you have any feature in oslo libraries, make sure it gets merged before Feb 1614:25
whoami-rajatthat's all the announcements we had for today14:26
whoami-rajatanyone has anything else?14:26
whoami-rajatokay, let's move to topics then14:28
whoami-rajat#topic Improvement to cinder-backup dealing with incremental backups14:28
whoami-rajatcrohmann, that's you14:28
crohmannYes. Sorry for hijacking this as a "topic". But I just wanted some quick feedback if you like the idea at all.14:28
zaitcevNo, it's fine.14:29
crohmannThis is a change by someone else, removing the fetching of ALL backups to then loop over them to find the parent.14:29
crohmannI reworked it and started to introduce a helper to just do a SINGLE DB query to find the suitable parent backup. I just need some guidance on using either BackupLists from objects to put the ujelp ... 14:30
whoami-rajati can see it's a very old patch from 2017 ...14:30
crohmannYeah. But we are heavy users of incremental backups, that's why I found and refactored it.14:31
whoami-rajatthe idea makes sense to me, need to check the approach but should be good to include14:31
crohmannThe loop and all of this "in Python" logic seemed messy.14:31
crohmannJust please someone use Gerrit to give me some feedback on where to place the helper funktion best and all. I finish the change quickly then.14:32
crohmannThat's all I wanted to ask about this one.14:34
zaitcevI'll take a look. But you might want to make sure eharney looks at the RBD one.14:34
crohmannRBD one ?14:35
zaitcevCeph has its own backup, kept in RBD. That is what their driver does.14:37
crohmannCan you drop a link to the change you are referring to?14:37
crohmannThe change I am discussing here is just about the cinder-backup DB objects. Nothing driver specific ther.14:38
zaitcevhttps://review.opendev.org/c/openstack/cinder/+/81045714:39
crohmannahhh .. you are already talking about the next next topic ;-) 14:40
eharneyon 484729, openstack-tox-pep8 passed but definitely should have failed -- weird.  wonder what's going on there14:40
whoami-rajat#link https://review.opendev.org/c/openstack/cinder/+/48472914:41
crohmannK, so zaitcev and eharney kindly look into https://review.opendev.org/c/openstack/cinder/+/484729 :-)14:42
crohmannwhoami-rajat: Which topic are we then on currently.14:42
whoami-rajatcrohmann, i think you have 3 topics, guess we are done with the first one?14:42
crohmannmore than, yes.14:43
whoami-rajat#topic DB indexes missing in backups table14:43
whoami-rajat#link https://bugs.launchpad.net/cinder/+bug/204839614:43
crohmannAlso I quick one... This is a bug about the backups table not having any indexes at all. But there are likely some columns that are used quite heavily. With Caracal coming up, this is the chance to add then.14:44
eharneywe should fix this, i'm interested in working on it  (took the bug for now)14:44
whoami-rajati remember geguileo did some work on indexes for some tables, but maybe backups was not a part of that effort14:45
whoami-rajatgreat, thanks eharney 14:45
crohmanneharney: Please also consider https://review.opendev.org/c/openstack/cinder/+/484729 which will data_timestamp :-) SCNR.14:45
crohmannI meant to say: "will filter on data_timestamp column"14:46
whoami-rajatso this looks addressed, anything else on this topic?14:46
crohmannnot from me. Just did not want to miss the "C" release for this schema change ;.)14:47
whoami-rajatok14:47
whoami-rajat#topic Cinder-Backup (rbd driver): add option to keep only last n snapshots per backup14:47
whoami-rajat#link https://review.opendev.org/c/openstack/cinder/+/81045714:47
crohmannThat's the one zaitcev mentioned earlier. eharney was called out to look into this one.14:48
whoami-rajatokay great14:48
whoami-rajatso last one from crohmann 14:49
whoami-rajat#topic Any update on the Ceph auth / caps disucssion?14:49
crohmannReview (and merge) of this feature kinda stalled. With a new comment from another operator today I got my hopes up we could get this14:49
crohmannin to Caracal14:49
whoami-rajatwe can surely look for reviews on that14:49
crohmannbut the last update was Jun 22, 2022. Would be awesome to revive the review. Let me / Jan know if there are any change required to get this merged.14:51
whoami-rajatsure, sorry it is taking long but we more or less have review bandwidth issues with contributors joining/leaving, will look more into that general problem14:54
whoami-rajatokay anything else on this? I'm planning to cover the next 2 topics quickly in next 5 minutes14:55
crohmannnope. I'm good on this one.14:56
whoami-rajatgreat, thanks14:56
whoami-rajat#topic Third party CI compliance check14:56
whoami-rajat#link https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/QK223RS27RYNDACTPUVJZR2REL42OC4B/14:56
whoami-rajatI've sent out a mail to ML so vendors please take a look14:56
whoami-rajatIf you are a vendor, please take a look and perform required actions before the midcycle meeting next week14:56
whoami-rajatthat's all i wanted to say14:56
whoami-rajat#topic volume type extra_specs behavior14:56
Luzisetting custom extra specs for volume types work, but creating a volume from it will fail during scheduling due to the need to satisfy ALL extra_specs in the backend. This works different from other resources like volumes themself. What is the desired state here? Allowing "volume type set --property customkey=customvalue" and letting the scheduler ignore this, or not allowing to set custom properties at all?14:56
whoami-rajat#link https://lists.openstack.org/archives/list/openstack-discuss@lists.openstack.org/thread/ZWXFYIQ3FSFC5MGTSIMVEHCEJRQQRQ3X/14:56
whoami-rajatwith custom properties, we are talking about driver specific properties right?14:57
eharneyi think it would be helpful to frame what you're wanting to accomplish with this14:58
whoami-rajatwe also allow setting AZs and multiattach in the extra specs which are not related to drivers14:58
Luzii am talking about the field of "properties" or extra specs and its usage in volume types 14:58
whoami-rajatso there are some cinder specific properties, like AZ and multiattach as i mentioned, and some driver specific that are checked against the capabilities that the driver reports14:59
Luzibecause i would like to use something in volume types to give users more uniform information about volume types14:59
whoami-rajatas eharney said, it would be easier to help if we know about the use case you are trying to achieve14:59
eharneyso the use case is to try to have fields that are visible to users but not actually used for scheduling, makes sense15:00
Luziand it seems for other openstack resources, i could simply use the "properties"15:00
Luziyeah eharney 15:00
eharneycurrently that would just have to use the description field i guess15:01
whoami-rajatokay we are out of time15:02
whoami-rajatwe can continue the discussion next week15:02
whoami-rajator on the #openstack-cinder channel15:02
whoami-rajatplease take a look at the review requests15:02
whoami-rajatand next week we will have the midcycle-215:02
happystackerthks 15:02
whoami-rajatso see you all there15:02
whoami-rajatthanks everyone!15:02
crohmannThank you all!15:02
whoami-rajat#endmeeting15:02
opendevmeetMeeting ended Wed Feb  7 15:02:39 2024 UTC.  Information about MeetBot at http://wiki.debian.org/MeetBot . (v 0.1.4)15:02
opendevmeetMinutes:        https://meetings.opendev.org/meetings/cinder/2024/cinder.2024-02-07-14.00.html15:02
opendevmeetMinutes (text): https://meetings.opendev.org/meetings/cinder/2024/cinder.2024-02-07-14.00.txt15:02
opendevmeetLog:            https://meetings.opendev.org/meetings/cinder/2024/cinder.2024-02-07-14.00.log.html15:02
*** jbernard_ is now known as jbernard20:58

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