Wednesday, 2022-07-20

*** dviroel|out is now known as dviroel00:33
*** dviroel is now known as dviroel|out01:00
aneeeshp11Anyone using Software Factory for cinder/os-brick third party CI? I did install Software Factory and it is posting results to ci-sandbox project. Now planning to add the jobs for cinder. Could any one share playbooks or job yaml files that can be used for the cinder project?03:27
opendevreviewliuhuajie proposed openstack/cinder master: Remove None for get method, the default value is None
opendevreviewTushar Trambak Gite proposed openstack/cinder master: Delete volume and cloned-volume concurrently
*** geguileo is now known as Guest551805:57
opendevreviewRajat Dhasmana proposed openstack/cinder master: Add cinder-manage command to update service_uuid
opendevreviewTushar Trambak Gite proposed openstack/cinder master: Storage node configuration
opendevreviewMerged openstack/cinderlib stable/yoga: Don't use yum to install ceph
opendevreviewMerged openstack/cinderlib stable/yoga: Update deprecated zuul syntax
opendevreviewMerged openstack/cinderlib stable/xena: Don't use yum to install ceph
opendevreviewMerged openstack/cinderlib stable/xena: Update deprecated zuul syntax
opendevreviewMerged openstack/cinderlib stable/wallaby: Don't use yum to install ceph
opendevreviewMerged openstack/cinderlib stable/wallaby: Update deprecated zuul syntax
toskywhoami-rajat: the victoria patches for cinderlib are ready to be merged too (and after they merge, I can fix the ussuri and train ones)09:47
opendevreviewTushar Trambak Gite proposed openstack/cinder master: Storage node configuration
whoami-rajattosky, just wanted to get some clarity on the fixes, rosmaita wrote that one patch is fixing the gate and another is removing deprecated zuul syntax, just wanted to know till which branch they have an impact?09:51
toskythe one which fixes the deprecated zuul syntax is needed everywhere09:53
toskythe other is needed until the relevant job is fixed09:54
toskywhich is probably everywhere the ceph job uses ubuntu, because it was logically wrong to use the ansible yum task there (a mistake I did some time ago)09:54
*** dviroel|out is now known as dviroel11:34
opendevreviewOleg proposed openstack/cinder master: Add NVMe/TCP support to Dell EMC PowerStore driver
rosmaitawhoami-rajat: the yum->package fix needs to go back to ussuri (train is using centos-7, where using yum should be no problem)12:03
rosmaitathe zuul syntax needs to go as far back as we can get it12:03
whoami-rajatack, currently one of the victoria patch is waiting for another +2 and ussuri gate is failing12:03
rosmaitawhoami-rajat: it's weird, victoria is passing without having backported to it, but the ussuri failure looks like it needs that patch12:20
whoami-rajatTBH, cinderlib gate has been always weird for me12:22
rosmaitayeah, the trailing release model doesn't help12:24
whoami-rajatCinder meeting in #openstack-meeting-alt at 1400 UTC13:59
whoami-rajatjungleboyj rosmaita smcginnis tosky whoami-rajat m5z e0ne geguileo eharney walshh_ jbernard sfernand enriquetaso hemna fabiooliveira yuval tobias-urdin13:59
*** geguileor is now known as geguileo14:00
opendevreviewAlexander Deiter proposed openstack/cinder master: Fix Infinidat driver to return all iSCSI portals
opendevreviewEric Harney proposed openstack/cinder-tempest-plugin master: WIP: test multiattach volume works after disconnect
opendevreviewOleg proposed openstack/cinder master: Add NVMe/TCP support to Dell EMC PowerStore driver
opendevreviewEric Harney proposed openstack/cinder master: Tests: Check for CRLF terminators in pep8 env
simondodsleyeharney: this change will potentially all the current and new NVMe drivers so it needs to be resolved before driver merge freeze15:00
enriquetaso#startmeeting cinder_bs15:01
opendevmeetMeeting started Wed Jul 20 15:01:49 2022 UTC and is due to finish in 60 minutes.  The chair is enriquetaso. Information about MeetBot at
opendevmeetUseful Commands: #action #agreed #help #info #idea #link #topic #startvote.15:01
opendevmeetThe meeting name has been set to 'cinder_bs'15:01
enriquetasoWelcome back to the Cinder Bug Meeting15:02
enriquetasoCinder has 6 new bugs reported. Mostly low bugs:15:02
enriquetaso#topic Infinidat 15:02
enriquetasoCinder has two bugs regarding Infinidat and one of them it's kind of a big bug:15:02
enriquetaso"Infinidat Cinder driver multi-attach feature is broken"15:02
enriquetasoI think many drivers have made the same mistake in their multiattach implementations. This is a sign that we need a tempest test that actually tests multiattach15:03
rosmaitaenriquetaso: ++15:03
enriquetasoeharney ^15:03
eharneyi tossed up  but it probably doesn't work yet15:03
enriquetasothanks eric! 15:03
enriquetasoPlease keep an eye on this test so we can merge it as soon as possible.15:03
enriquetasomaybe infinidat team would like to add some comments to this ? :)15:04
enriquetasoThe second one is 15:05
enriquetaso"Infinidat Cinder driver ignores driver_use_ssl option"15:05
enriquetasoIt has a fix proposed to master, feel free to review it!15:05
eharneyinfinidat also has up which needs review, maybe someone looking at these reviews should hit all of them together15:06
enriquetasosounds good15:08
enriquetasoNext, we have 3 generic-NFS and remotefs related bugs:15:08
enriquetaso#topic  NFS Backup driver doesn't remove empty directories on backup deletion. 15:09
enriquetaso#link 15:09
enriquetasowhoami-rajat: Using cinder backup with NFS backend, everything works fine, except that when we delete a backup the actual "content" of the backup gets deleted but not the directory structure on the NFS share15:09
enriquetasoFix proposed to master:15:09
eharneyi called out some issues w/ race conditions on the patch, need to go look at rajat's updates and see if i can find any further concerns15:09
whoami-rajatyeah, I've revised with some of eharney's suggestions15:09
whoami-rajatthanks eharney 15:10
enriquetaso#topic [NFS] Nova raises an error on the server to resize command15:12
enriquetasoI have not been able to look into it more, so I am not sure if there is anything to do from the cinder side. I have a devstack+nfs so I'll try this out today.15:12
eharneyyeah, we need to determine if this is doing something that is not intended to be supported15:14
eharneya closer look would be good15:14
enriquetaso#action (enriquetaso) take a closer look and reproduce bug198156215:15
enriquetasoThe last one is incomplete and waiting from the reporter side but I'd like to mention it15:15
enriquetaso#topic RemoteFs: Unexcepted error while running the command.15:15
enriquetaso#link 15:15
eharneywe should be able to understand this one just by looking at the code15:16
eharneydoes the nfs driver call mkdir via rootwrap?  if so it needs to be in the volume.filters file15:16
enriquetasoOK, so it's not a incomplete bug15:18
enriquetasoit's a valid bug 15:18
eharneyi think only os-brick calls mkdir, so this is probably not the right fix, but unclear what is going on15:18
enriquetasoOK, maybe we can discuss it later in this week?15:19
eharneyit would repro on generic nfs, but may depend on how they have it configured15:20
eharneyobviously our gate job is setup in a way where mkdir works15:20
enriquetasoI'm not completely aware of what the fix want's to achieve, but since I have the nfs devstack maybe I can check this as well15:21
enriquetasomakes sense15:21
enriquetasothanks Eric15:21
eharneyyes, mkdir being run as root needs to be done, but the code that does that is in brick, and there is a privsep layer involved there instead of just rootwrap etc, need to peel back the onion a bit15:21
enriquetasoOK, moving one. The last bug of the day is from Storwize15:22
enriquetaso#topic [storwize] Driver Initialization error w.r.t default portset15:22
enriquetasoDriver initialization failed with the below error:15:23
enriquetaso ERROR cinder.volume.manager cinder.exception.InvalidInput: Invalid input received: The storage device does not support iSCSI. Please configure the device to support iSCSI or switch to a driver using a different protocol.15:23
enriquetasoThere's a fix proposed to master15:23
enriquetaso#topic open discussion15:24
enriquetasoPlease feel free to mention bugs now 15:24
eharney is one that could use some review15:24
enriquetasoPlease Cinder team review ^15:26
enriquetasoi'll do it 15:26
enriquetasoit's a medium size bug15:26
dwhite44There is a tempest bug affecting some CIs, maybe relevant if someone is digging into the multi-attach bug from earlier:
eharneyi don't know enough about tempest to really understand that bug, but good to know about15:30
enriquetasooh, thanks for sharing that dwhite44 i'll keep an eye on bug198081615:31
enriquetasothanks everyone!15:31
opendevmeetMeeting ended Wed Jul 20 15:31:46 2022 UTC.  Information about MeetBot at . (v 0.1.4)15:31
opendevmeetMinutes (text):
enriquetaso:P thanks15:31
whoami-rajatrosmaita, can you reply to the open concern here?
whoami-rajat 15:36
whoami-rajatthis comment specifically15:36
rosmaitawhoami-rajat: i am still thinking15:57
whoami-rajatok, I was also confused there15:58
rosmaitayou know, there have been no functional changes since 4.1.0, so we could just not release cinderlib xena15:59
rosmaitai'm not sure whether 4fc56c8 is for the gate only, or if it has wider implications15:59
rosmaitaso if that's all that's holding up the xena stable releases, i suggest drop the cinderlib.yaml and we can figure this out later16:01
*** dviroel is now known as dviroel|lunch16:06
whoami-rajatI'm not sure either about the impact of the patch16:18
whoami-rajatbut i tend to agree that the motive of the patch was to fix gate so not really needed in the release16:18
whoami-rajatjbernard, what do you think about it? ^16:19
jbernardwhoami-rajat: if it just addresses issue in the gate, it would be invisible to most (if not all) consumers of the stable release no?16:23
jbernardwhoami-rajat: if that's true, we can omit it from this round, imo16:24
whoami-rajatrosmaita, one ques, are we using cinderlib from source in other stable branches like cinder xena? or using the released version of cinderlib xena?16:30
opendevreviewAlexander Deiter proposed openstack/cinder master: Fix Infinidat driver to use SSL options
opendevreviewOleg proposed openstack/cinder master: Add NVMe/TCP support to Dell EMC PowerStore driver
opendevreviewOleg proposed openstack/cinder master: Add NVMe/TCP support to Dell EMC PowerStore driver
*** akekane_ is now known as abhishekk17:09
rosmaitawhoami-rajat: we only run cinderlib functional tests in cinder gates17:14
rosmaitacinderlib gates are from source (as you'd expect)17:15
rosmaitabut to answer your question, looks like the cinder CI uses cinderlib from source17:17
rosmaitaafaik, the only cinderlib job affected by that change is cinderlib-lvm-functional (which for some reason is passing on victoria without the patch, whereas ussuri is failing: )17:19
*** dviroel|lunch is now known as dviroel17:23
whoami-rajatso the cinderlib release doesn't affect gate situation in other projects17:43
whoami-rajatrosmaita, jbernard i think we should remove cinderlib from xena releases then 17:44
whoami-rajatrosmaita, also regarding the failure, backporting the cinderlib fix to ussuri should fix it ?17:44
jbernardwhoami-rajat: ok, ill update teh patch17:45
jbernardwhoami-rajat: done17:48
jbernardwhoami-rajat: that one should be good to go now17:48
rosmaitawhoami-rajat: i think so ... the fix is in wallaby, somehow victoria doesn't need it, but i'm pretty sure it's exactly the problem we are seeing in ussuri17:51
rosmaitai have been poking around trying to figure out why victoria doesn't seem to need it, can't figure anything17:52
rosmaitawhoami-rajat: i will propose backports to victoria and wallaby, because i don't think we want to skip victoria with that change?17:54
jbernardwhoami-rajat: i mentioned to elod that you would circle back on his stable-xena questions17:55
jbernardwhoami-rajat: (i didn't mean to overwrite them)17:55
opendevreviewEric Harney proposed openstack/cinder master: RBD: Flattening of child volumes during deletion
opendevreviewEric Harney proposed openstack/cinder master: Tests: RBD: Refactor mocks
eharneywhoami-rajat: geguileo: rosmaita: we merged w/ the cinder-mypy job failing, so cinder-mypy now fails on master17:59
whoami-rajatrosmaita, yep, don't want to miss a release even though we're not releasing it (i.e. victoria)18:15
whoami-rajatjbernard, ack, will leave a reply there18:15
whoami-rajateharney, ah, sorry about that, i will propose a fix for it18:16
opendevreviewRajat Dhasmana proposed openstack/cinder master: Fix mypy job
whoami-rajateharney, rosmaita fixed it ^18:53
rosmaitawhoami-rajat: good job!18:55
rosmaitawhoami-rajat: turns out that wasn't introduced until wallaby, so we can't cherry-pick directly to victoria18:56
rosmaitatrain, ussuri, and victoria are all in Extended Maintenance mode anyway at this point18:56
rosmaitaif no more work is going to go into those, we could remove the affected job and EOL those branches18:56
rosmaitawe'll have to see what geguileo thinks ... i'm not sure what the user community for cinderlib is like18:57
rosmaitahopefully, they like to use the latest release18:57
whoami-rajatah so if we want to fix it, we will need the privsep fix + this patch which introduced privsep, strange to see those branches failing then if it doesn't even support privsep18:59
whoami-rajats/support privsep/ have privsep support19:00
whoami-rajatI'm not sure either but sooner or later they will be EOL so we should decide if we want to maintain it or not19:00
whoami-rajatwow, even Pike is EM till now ...
rosmaitapike is eol for cinder, anyway19:32
*** dviroel is now known as dviroel|afk19:34
opendevreviewSimon Dodsley proposed openstack/cinder master: Add additional transport type constants
opendevreviewSimon Dodsley proposed openstack/cinder master: Add Pure Storage NVMe-RoCE driver
opendevreviewSimon Dodsley proposed openstack/cinder master: Add additional transport type constants
opendevreviewSimon Dodsley proposed openstack/cinder master: Add additional transport type constants

Generated by 2.17.3 by Marius Gedminas - find it at!