14:00:14 <rosmaita> #startmeeting cinder
14:00:15 <openstack> Meeting started Wed Jan 22 14:00:14 2020 UTC and is due to finish in 60 minutes.  The chair is rosmaita. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:00:16 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:00:18 <openstack> The meeting name has been set to 'cinder'
14:00:22 <smcginnis> o/
14:00:22 <e0ne> hi
14:00:29 <rosmaita> #link https://etherpad.openstack.org/p/cinder-ussuri-meetings
14:00:38 <whoami-rajat> hi
14:00:39 <rosmaita> #topic roll call
14:00:41 <geguileo> hi! o/
14:01:22 <m5z> hi
14:02:00 <tosky> o/
14:02:08 <enriquetaso> o/
14:02:19 <rosmaita> looks like a good turnout
14:02:30 <rosmaita> #topic announcements
14:02:42 <rosmaita> spec freeze is next week
14:03:07 <rosmaita> thanks to everyone who attended the virtual mid-cycle yesterday, i think it was productive
14:03:19 <rosmaita> #link https://wiki.openstack.org/wiki/CinderUssuriMidCycleSummary
14:03:29 <rosmaita> ^^ that's a quick writeup of what we talked about
14:03:30 <enriquetaso> rosmaita++ thanks to you!
14:03:52 * rosmaita blushes
14:04:13 <rosmaita> i put together a survey for participants but also people who did NOT attend
14:04:24 <rosmaita> #link https://forms.gle/muR2vMArZ7Eu1iV46
14:04:38 <rosmaita> will help in planning for part 2, which will be week of 16 March
14:04:58 <enriquetaso> \o/
14:05:05 <rosmaita> and, i have been a busy boy, here's the result of one of my action items:
14:05:15 <rosmaita> #link https://review.opendev.org/703709
14:05:18 <e0ne> rosmaita: awesome! I like such format with several short calls
14:05:35 <rosmaita> e0ne: that's good feedback
14:06:06 <rosmaita> the above link is an attempt to be explicit about how we want to write/review code that uses py3-only features, and also code that still uses py2 compatibility
14:06:17 <rosmaita> please be brutal in your comments!
14:06:33 <rosmaita> it will be helpful if we can get some good guidelines together
14:06:53 <e0ne> rosmaita: thanks for this patch! I didn't get a time to review it but it's great to have such guides
14:07:09 <rosmaita> thank you
14:07:18 <rosmaita> ok, that's all for announcements, on to the main program
14:07:27 <rosmaita> #topic really quick spec review
14:07:40 <rosmaita> #link https://review.opendev.org/#/q/status:open+project:openstack/cinder-specs
14:08:04 <rosmaita> i just want to take a quick look at these, mainly asking people if we are still interested
14:08:15 <rosmaita> or, if they are still interested in proposing them
14:08:28 <smcginnis> I don't think the backup-id one needs to be a spec. We should just add the code.
14:08:46 <rosmaita> yeah, i have a question about it later on the agenda
14:09:10 <rosmaita> i was going to say, we could leave it as a spec for when this comes up again about adding something to the volumes table
14:09:11 <rosmaita> but
14:09:20 <rosmaita> we can point people to this review, it doesn't have to be a spec
14:09:24 <whoami-rajat> i think we can merge 'copy image to multiple stores' patch if everyone is ok with it (looks like it from the +2 list)
14:09:41 <rosmaita> yes, i had a question about that
14:09:50 <rosmaita> is it cinder policy for the PTL to merge specs?
14:09:59 <rosmaita> guess that's for smcginnis and jungleboyj
14:10:04 <rosmaita> the question, i mean
14:10:19 <smcginnis> rosmaita: I don't think we ever made it a formal policy.
14:10:23 <jungleboyj> o/
14:10:31 <jungleboyj> Sorry.  Late.
14:10:42 <rosmaita> i definitely like the idea of as many cores reading each spec as possible
14:11:01 <whoami-rajat> rosmaita, ++
14:11:16 <rosmaita> so it's probably good to leave them open even with multiple +2s
14:11:29 <rosmaita> but i agree with whoami-rajat that this particular spec is ready for merging
14:11:48 <rosmaita> i didn't see eharney come in
14:11:49 <e0ne> I didn'
14:12:03 <e0ne> I didn't approve this spec because it was added to the agenda
14:12:39 <rosmaita> e0ne: you mean the multiple stores of glance spec?
14:12:47 <rosmaita> did you still want to look it over?
14:13:07 <e0ne> rosmaita: I was talking about https://review.opendev.org/#/c/700977
14:13:30 <rosmaita> e0ne: ok
14:14:03 <rosmaita> i was going to ask eharney if he plans to revise https://review.opendev.org/682456 , i can ask offline
14:14:26 <rosmaita> it looks like everything else has been sitting for over a month
14:14:49 <rosmaita> so it looks like we are pretty much on schedule for specs
14:15:11 <rosmaita> although, Liang did have a question about the volume-local-cache spec
14:15:48 <rosmaita> from yesterday's discussion, it looked like maybe cinder needs to know about what cache-mode is available so we can reject unsafe modes?
14:15:59 <rosmaita> or did I misunderstand what we wanted there?
14:16:34 <rosmaita> Liang's question was how we would get this info to cinder
14:16:46 <rosmaita> i guess the issue is:
14:17:08 <rosmaita> if we want to disallow unsafe modes completely, we can have os-brick reject the attachment
14:17:26 <rosmaita> if we want to disallow unsafe operations, we would need cinder to have some info
14:18:16 <rosmaita> my request is that if you have an interest in this spec, and especially in this part of it, please leave a comment on the spec
14:18:53 <rosmaita> i think i may have gone over my 5 minutes
14:19:05 <rosmaita> #topic 3rd-party CI update
14:19:09 <rosmaita> e0ne: that's you
14:19:17 <e0ne> rosmaita: thanks
14:19:52 <e0ne> I'm trying to figure out how many 3rd-party CIs have cinder tempest plugin running
14:20:18 <e0ne> I tried to check on existing patches to cinder but a lot of logs are not available now
14:20:31 <e0ne> that's why I created a dummy patch
14:20:43 <e0ne> #link https://review.opendev.org/#/c/703772/
14:20:51 <rosmaita> good idea
14:20:52 <jungleboyj> e0ne:  :-)  Yeah, that is part of the reason I haven't had a chance to look at all of that.
14:21:00 <e0ne> and I tried ti run all CIs from https://docs.openstack.org/cinder/latest/drivers.html
14:21:01 <e0ne> #link https://docs.openstack.org/cinder/latest/drivers.html
14:21:19 <e0ne> there are a lot of failures
14:21:29 <e0ne> and I'm waiting for more results
14:21:34 <jungleboyj> :-(
14:21:43 <e0ne> probably, I'll to it on a daily basis to run rechecks
14:21:47 <e0ne> https://docs.google.com/spreadsheets/d/1id4FY7Ywv8vWp01cTw_Km_YpN98U7vYvFgn8wy_o9vU/edit#gid=0
14:21:53 <e0ne> I'm sorry for google docs
14:22:06 <e0ne> I can convert it to some more open format if needed
14:22:23 <rosmaita> i think that's fine for now
14:22:26 <e0ne> I'm worried a lot about Dell EMC CI
14:22:41 <smcginnis> I did get copied on a message that the log server is down.
14:22:46 <e0ne> I was not able to open any logs from it both from Europe and US
14:22:53 <jungleboyj> That is fine.  Ugh, what is up with the Dell CI not having logs accessible?
14:22:54 <smcginnis> There is an IT ticket open to resolve that.
14:23:01 <tosky> so, to summarize, and without considering the Dell CI ...
14:23:07 <jungleboyj> smcginnis:  Ah, ok.  Thanks for being on that.
14:23:11 <e0ne> smcginnis: thanks for the information
14:23:16 <tosky> just one 3rd party CI is running cinder-tempest-plugin (!)
14:23:28 <e0ne> Fujitsu ETERNUS CI logs are not accessible too:(
14:23:30 <rosmaita> wow
14:23:46 <jungleboyj> Are we really down to this few drivers?
14:23:49 <rosmaita> e0ne: yes thanks for putting this all in one place
14:23:49 <e0ne> it reports SUCCESS
14:23:54 <e0ne> but there are no logs http://openstackci.jp.fujitsu.com/Eternusci/72/703772/1/check/fujitsu-eternus-dx-iscsi/6d794b6
14:23:59 <jungleboyj> rosmaita: ++
14:24:14 <e0ne> ^^ this job reported to gerrit 25 minutes ago
14:24:46 <e0ne> my google spreadsheet is open for comments for everybody
14:25:11 <e0ne> I can give add write permissions per request
14:25:45 <e0ne> according tempest plugin, only NetApp SolidFire CI is using it:(
14:26:03 <e0ne> Datara CI should use it too but I don't have a confirmation
14:26:05 <rosmaita> well, i guess we should say "Hooray for NetApp!"
14:26:13 <e0ne> I
14:26:21 <rosmaita> but otherwise, this is a depressing spreadsheet
14:26:21 <jungleboyj> :-)
14:26:29 <e0ne> I'll keep this doc updated once I'll have more results
14:26:46 <tosky> thanks; so what are the next steps?
14:26:47 <e0ne> Pure Storage CI has own tempest plugin with few tests
14:27:04 <tosky> e0ne: do you know if that plugins is published somewhere?
14:27:14 <tosky> I wonder if those tests are really pure-specific
14:27:22 <e0ne> tosky: I don'd know yet
14:27:28 <tosky> anyway, do we have direct contacts for all CI maintainers?
14:27:36 <e0ne> unfortunately, that's all I've got for today:(
14:27:51 <smcginnis> They *should* all be listed on
14:28:01 <rosmaita> e0ne: that's plenty! thanks for putting this together
14:28:01 <smcginnis> https://wiki.openstack.org/wiki/ThirdPartySystems
14:28:03 <e0ne> https://wiki.openstack.org/wiki/ThirdPartySystems/Dell_EMC_SC_Series_CI
14:28:20 <e0ne> you can find CI wiki with contacts on the https://docs.openstack.org/cinder/latest/drivers.html page
14:28:30 <e0ne> tosky: ^^
14:28:41 <jungleboyj> tosky: In the past I have used the infor smcginnis shared.
14:28:55 <jungleboyj> I get some responses.  Sometimes I don't.
14:28:59 <rosmaita> i think we may need to have an emergency virtual mid-cycle focusing only on 3rd party CI
14:29:20 <e0ne> +1
14:29:22 <jungleboyj> Yeah.  :-(
14:29:44 <e0ne> btw, I didn't check if 3rd-party CI uses python3
14:29:53 <rosmaita> jungleboyj: i think you took an action item yesterday to contact some people who's CIs are not responding?
14:30:32 <jungleboyj> rosmaita:  Yes I did.  Will try to follow up on that today.
14:30:41 <rosmaita> ok, great
14:30:53 <jungleboyj> As well as making a blanket statement to the ML to see if we get more discussion.
14:31:08 <rosmaita> let's discuss this again at next week's meeting
14:31:24 <jungleboyj> rosmaita: ++
14:31:29 <whoami-rajat> e0ne, i think the CI's not using python3 were marked unsupported in Train
14:31:46 <e0ne> whoami-rajat: sounds reasonable
14:31:51 <rosmaita> yes, hopefully no one snuck back to using py2!
14:32:13 <jungleboyj> whoami-rajat: Right.  I went through that process.
14:33:04 <rosmaita> ok, let's revisit this next week and in the meantime I'll follow up with something more concrete
14:33:15 <rosmaita> (not sure what, though)
14:33:19 <e0ne> :)
14:33:52 <rosmaita> in the meantime, please look at this from yesterday: https://wiki.openstack.org/wiki/CinderUssuriMidCycleSummary#The_current_state_of_3rd_Party_CIs
14:33:58 <rosmaita> #link https://wiki.openstack.org/wiki/CinderUssuriMidCycleSummary#The_current_state_of_3rd_Party_CIs
14:34:03 <jungleboyj> Well, we have been going back and forth on this forever.  Who knows, you might have a new insight rosmaita
14:34:12 <rosmaita> one never knows
14:34:33 <rosmaita> yeah, smcginnis posted some info from 2017 on this same topic yesterday!
14:35:08 <rosmaita> thanks for following up on this e0ne
14:35:15 <rosmaita> any more comments?
14:35:20 <e0ne> rosmaita: you're welcome
14:35:27 <e0ne> nothing from my side for now
14:35:38 <smcginnis> That definitely wasn't the first time it was discussed either. ;)
14:35:46 <rosmaita> right!
14:36:18 <jungleboyj> And won't be the last.
14:36:25 <rosmaita> #topic is 'backup_id' in volume metadata ambiguous?
14:36:34 <rosmaita> actually, before i pose that question
14:36:48 <rosmaita> i think smcginnis has a good point that the spec should be a bug instead
14:37:00 <rosmaita> #link https://review.opendev.org/#/c/700977
14:37:11 <rosmaita> ^^ so you know what i'm talking about
14:37:34 <smcginnis> It's metadata. We have a field for metadata. It's not captured in there yet. Just add it (my take on things)
14:37:55 <rosmaita> that makes sense to me
14:38:15 <smcginnis> There's no real API change, no need to modify the DB, so not really anything that needs to be captured in a spec to make sure everyone understands the change and impact.
14:38:40 <rosmaita> and there's the backport potential you mentioned on the review
14:38:58 <smcginnis> Yeah, I think there's a good case for that.
14:39:03 <rosmaita> we can leave it up to the spec proposers about how they want to follow up
14:39:07 <rosmaita> i'll leave a comment
14:39:09 <rosmaita> in the meantime
14:39:27 <rosmaita> my question is whether 'backup_id' in the volume metadata could be ambiguous
14:39:38 <rosmaita> is it obvious that it means "This volume was created from a backup" vs. "This volume was backed up"?
14:39:50 <smcginnis> Maybe "src_backup_id"?
14:39:52 <rosmaita> i'm kind of new around here, that's why i'm asking
14:40:01 <rosmaita> smcginnis: yes, that's what i was thinking
14:40:12 <smcginnis> I like that it is more explicit then.
14:40:36 <rosmaita> ok, i will suggest that on the spec
14:41:01 <smcginnis> The spec also focuses on volume creation directly from backup, but that was just a convenience API added. I think the change should also handle adding that metadata when a volume is independently created, then restored to from a backup.
14:41:08 <enriquetaso> smcginnis++
14:41:36 <rosmaita> i agree, might as well make it as useful as possible
14:41:56 <rosmaita> any other comments?
14:42:53 <smcginnis> Nope
14:42:59 <rosmaita> ok, moving on
14:42:59 <rosmaita> #topic py3.6 *and* py3.7 testing in the gate
14:43:12 <rosmaita> this is another follow-up from yesterday
14:43:39 <rosmaita> since we don't have to test py2 any more, and we are claiming py3.6 and py3.7 support, i figure we should be testing both of those
14:43:52 <rosmaita> (it won't consume any extra resources)
14:43:59 <e0ne> +1
14:44:11 <rosmaita> anyway, i went back and looked at our recent change to tox.ini and .zuul.yaml
14:44:18 <rosmaita> results are here:
14:44:28 <rosmaita> #link https://etherpad.openstack.org/p/cinder-ussuri-community-goal-drop-py27-support
14:44:40 <rosmaita> down at the bottom, under "Review"
14:45:11 <smcginnis> 3.8 may be added in Victoria. There is some discussion on that that just testing 3.6 and 3.8 should be sufficient to make sure 3.7 is safe.
14:45:19 <smcginnis> Just a data point for the future.
14:45:30 <rosmaita> ok, good to know
14:45:47 <rosmaita> our current setup will be easy to add 3.8 testing to both tox and zuul
14:46:31 <rosmaita> in any case, i think cinderlib and os-brick are running gate tests with just the system python3 for wherever the test lands
14:46:37 <rosmaita> which may be ok
14:46:47 <rosmaita> but i wanted to point it out
14:47:04 <rosmaita> as far as cinder-tempest-plugin ...
14:47:17 <rosmaita> looks like it's running the normal tests plugins are supposed to run
14:47:29 <rosmaita> (although I guess I should verify that)
14:47:55 <rosmaita> so i'm inclined not to mess with it
14:48:12 <whoami-rajat> with python3 right?
14:48:24 <rosmaita> whoami-rajat: yes
14:48:39 <whoami-rajat> ok
14:48:40 <smcginnis> I think I saw some work going on with tempest itself about how to handle things since tempest is branchless.
14:48:42 <rosmaita> which is most likely py 3.6, i think
14:49:05 <rosmaita> yes, smcginnis you have a good point, we should let that shake out before making changes ourselves
14:49:20 <smcginnis> I think it may have just required some devstack changes to allow stable branches to run services with py2.7 but run tempest under py3.
14:50:39 <rosmaita> so, to summarize, it looks like our testing is in pretty good shape
14:50:57 <rosmaita> could be some room to look at cinderlib and os-brick if anyone is so inclined
14:51:21 <rosmaita> but those gate jobs are pretty complicated, i personally do not want to touch them with a 10-foot pole
14:51:32 <smcginnis> ;)
14:51:51 <jungleboyj> /me hands rosmaita  an 11 foot pole
14:52:05 <rosmaita> jungleboyj: i thought maybe you would go metric!
14:52:16 <jungleboyj> :-)
14:52:31 <rosmaita> i guess it's a 3 metre poll for our non-USA constituents
14:52:42 <rosmaita> ok, enough of that
14:52:47 <rosmaita> #topic open discussion
14:53:20 <lseki> sorry I'm late
14:53:23 <lseki> e0ne: netapp ontap drivers {nfs, iscsi, fcp} are running cinder-tempest-plugin as well
14:53:27 <lseki> we're having some troubles with our CI right now, though
14:53:33 <lseki> I'll let you know when it's fixed and having plugin tests passing
14:53:39 <e0ne> lseki: thanks
14:53:41 <rosmaita> lseki: ty
14:53:44 <jungleboyj> lseki:  ++  Great.
14:53:56 <rosmaita> lseki: you missed our shout-out to netapp for using cinder-tempest-plugin
14:54:06 <lseki> \o/
14:54:11 <jungleboyj> whoop whoop!
14:54:20 <rosmaita> (09:26:05 AM) rosmaita: well, i guess we should say "Hooray for NetApp!"
14:54:24 <lseki> hooray 🎉
14:54:31 <dviroel> \o/
14:54:36 <enriquetaso> hooray :P
14:54:48 <sfernand> :)
14:55:41 <smcginnis> Oh, since it's open floor, just pointing out the naming for W - https://wiki.openstack.org/wiki/Release_Naming/W_Proposals
14:56:09 <rosmaita> so anything goes nowadays, naming wise?
14:56:32 <jungleboyj> This will be interesting.
14:56:33 <smcginnis> Yep.
14:56:50 <smcginnis> Up to 10 letters long in the ISO basic Latin alphabet.
14:57:08 <rosmaita> well, i am all out of ideas after "Vaiveahtoish"
14:57:17 <smcginnis> :)
14:57:20 <rosmaita> (which obviously didn't win)
14:58:36 <rosmaita> down to 2 minutes -- anyone?
14:59:05 <whoami-rajat> a review request for the patch - https://review.opendev.org/#/c/700176/
14:59:07 <LiangFang> o/
15:00:14 <rosmaita> LiangFang: noted
15:00:32 <rosmaita> ok, thanks to e0ne for his 3rd party CI spreadsheet
15:00:40 <rosmaita> and see everyone next week!
15:00:44 <rosmaita> #endmeeting