16:00:03 <smcginnis> #startmeeting cinder
16:00:03 <openstack> Meeting started Wed Mar 20 16:00:03 2019 UTC and is due to finish in 60 minutes.  The chair is smcginnis. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:05 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:00:07 <openstack> The meeting name has been set to 'cinder'
16:00:09 <whoami-rajat> Hi
16:00:24 <smcginnis> Courtesy ping: diablo_rojo, diablo_rojo_phon, rajinir tbarron xyang xyang1 e0ne gouthamr thingee erlon tpsilva ganso patrickeast tommylikehu eharney geguileo smcginnis lhx_ lhx__ aspiers jgriffith moshele hwalsh felipemonteiro lpetrut lseki _alastor_ whoami-rajat yikun rosmaita enriquetaso hemna _hemna
16:00:32 <geguileo> hi! o/
16:00:34 <lseki> hi
16:00:35 <eharney> hi
16:00:37 <abishop> o/
16:00:42 <rosmaita> o/
16:00:47 <xyang> Hi
16:00:48 <e0ne> hi
16:00:57 <smcginnis> Jay has some personal business to attend to today, so I'm your guest moderator for today. :)
16:01:09 <smcginnis> #link https://etherpad.openstack.org/p/cinder-stein-meetings Agenda
16:01:14 <geguileo> smcginnis: thanks
16:01:22 <whoami-rajat> :( :)
16:02:07 <smcginnis> Guess we have a pretty good turnout, so I'll get going.
16:02:11 <smcginnis> #topic Announcements
16:02:27 <smcginnis> Wanted to make sure folks were aware tomorrow is the RC1 deadline.
16:02:29 <smcginnis> #link https://releases.openstack.org/stein/schedule.html
16:02:35 <smcginnis> We are getting really close to the end of stein now.
16:02:53 <e0ne> there is also Hard String Freeze deadline
16:03:05 <smcginnis> e0ne: Oh, good point.
16:03:14 <smcginnis> There are multiple things that come with this deadline.
16:03:35 <whoami-rajat> smcginnis: so my api-ref patches won't merging after it? right?
16:03:59 <smcginnis> whoami-rajat: Probably not for stein.
16:04:15 <e0ne> after RC1 master branch will be opened for Train, so any fixes to Stain should be backported to stable/stien
16:04:17 <smcginnis> But technically, as soon as we cut RC1 and a stable/stein branch, that makes master Train.
16:04:29 <diablo_rojo_phon> Also if cinder has cycle highlights.. getting those in ASAP.
16:04:48 <smcginnis> But we should still be focusing on stein work until the cycle is done.
16:04:55 <smcginnis> diablo_rojo_phon: Thanks!
16:05:02 <e0ne> smcginnis: +1
16:05:10 <smcginnis> With Jay tied up, I guess I'll try to put together some highlights.
16:05:20 <smcginnis> Could use input from others on that though.
16:05:34 <diablo_rojo_phon> smcginnis: thank you for the convenient segue :)
16:05:51 <diablo_rojo_phon> And in advance for the highlights.
16:05:55 <jungleboyj> smcginnis: I have that on my list.
16:06:06 <smcginnis> jungleboyj: Oh, awesome
16:06:11 <jungleboyj> I will get them done.
16:06:30 <jungleboyj> diablo_rojo_phon: thanks for thr reminder.
16:06:38 <smcginnis> All cores, please mark patches as review-priority +1 or +2 if you think there is anything critical to get in for stein yet.
16:07:15 <smcginnis> The Train schedule has also been published.
16:07:18 <smcginnis> #link https://releases.openstack.org/train/schedule.html
16:07:32 <diablo_rojo_phon> jungleboyj: if you have any questions myself or other release folks can help. I have a meeting Thursday to start processing all the info so by EOD today would be preferable if possible?
16:08:18 <jungleboyj> diablo_rojo_phon: ok. Will do. Have done those before. Just need to figure out what to say.
16:08:44 <smcginnis> diablo_rojo_phon: I'm thinking those kinds of meetings either need to be held later, or we should move the highlights deadline up to be earlier in train.
16:09:08 <diablo_rojo_phon> smcginnis: I think both :)
16:09:33 <jungleboyj> smcginnis: ++
16:09:34 <diablo_rojo_phon> jungleboyj: happy to help if you need me.
16:10:01 <jungleboyj> diablo_rojo_phon: cool thanks.
16:10:02 <diablo_rojo_phon> I would have pushed the meeting later if I could. It was slated to happen last week.
16:10:12 <smcginnis> One more announcement - PTL election for Train is over. Long live the king (of the jungle).
16:10:28 <jungleboyj> Ha ha.
16:11:01 <smcginnis> #topic Generalized filtering fixes
16:11:04 <smcginnis> #link https://review.openstack.org/#/c/641717/
16:11:09 <enriquetaso> o/
16:11:24 <smcginnis> I wanted to bring this one up because I think we want to get this in stein.
16:11:31 <smcginnis> But I've also not been following things as closely.
16:11:37 <eharney> yeah, we should get it in if we can
16:11:38 <smcginnis> So hoping for input from whoami-rajat and eharney
16:12:02 <smcginnis> eharney did mark it review-priority +1, so assuming so. So we also need some eyes on it.
16:12:43 <smcginnis> eharney: I think there was some question at some point about the correct approach to fix things. Is this where we want it now?
16:12:58 <smcginnis> eharney: You didn't +2 it, so wasn't sure if there were still questions.
16:13:19 <whoami-rajat> I've been finding different cases and the ones i tested i've added so far, if anyone faces problems in any command as a non-admin user, they can notify me so i can append into the patch.
16:13:27 <eharney> smcginnis: it's the right approach -- the tricky part was to review it thoroughly enough to make sure there weren't any unexpected side effects, mostly around knowing that all of the correct checking exists around "all_tenants" for each resource etc
16:13:28 <smcginnis> whoami-rajat: I see eharney pointed out a bug it should be linked to. Can we get that updated quick?
16:14:15 <whoami-rajat> smcginnis: sure will do.
16:15:18 <smcginnis> OK, I'll try to review later today, but call for help from other cores to check that out.
16:15:42 <whoami-rajat> smcginnis: done
16:15:45 <smcginnis> #topic RSD driver follow up
16:15:47 <smcginnis> whoami-rajat: THanks
16:15:57 <smcginnis> tejdeep: All yours.
16:16:13 <smcginnis> I see this was on the agenda last week. Was there some follow up to cover?
16:16:32 <smcginnis> #link https://review.openstack.org/621465
16:17:00 <smcginnis> tejdeep: You around?
16:17:35 <smcginnis> Well, I'll move on since I'm not sure why we're covering new drivers in the meeting at this point anyway.
16:17:42 <smcginnis> #topic cinderlib followup
16:18:00 <smcginnis> geguileo: You have the floor.
16:18:02 <geguileo> thanks
16:18:09 <geguileo> so this is a follow up of last week patches
16:18:26 <geguileo> there's been some reviews and iterations, but they are still pending
16:18:38 <geguileo> but the most important one is the one that fixes the package
16:18:46 <geguileo> which is currently broken (my bad)
16:19:02 <geguileo> so now I've split the publish to pypi patch in 2
16:19:08 <geguileo> the fix of the package in cinderlib
16:19:18 <geguileo> #link https://review.openstack.org/643542
16:19:36 <geguileo> sorry, that's the pypi job that's now in project-config
16:19:47 <geguileo> this is the one that fixes cinderlib: https://review.openstack.org/643013
16:19:58 <geguileo> it's important to get ^ merged
16:20:15 <geguileo> without it we cannot built tarballs
16:20:18 <smcginnis> Added to my review queue.
16:20:23 <geguileo> smcginnis: thanks!!!
16:20:32 <tejdeep_> hi sorry i lost connection while giving update
16:20:40 <geguileo> and the distros need the tarballs to test their packages for Stein release
16:21:20 <smcginnis> Yeah, we've merged most of it, so we should follow through and get it in good shape for stein.
16:21:36 <geguileo> that would be awesome
16:22:17 <geguileo> that's all I had to say
16:22:24 <smcginnis> OK, thanks geguileo
16:22:28 <smcginnis> #topic RSD driver
16:22:37 <smcginnis> tejdeep_: Was there some follow up from last week?
16:23:19 <tejdeep_> nothing, CI status looks good, just following up on merge status
16:24:21 <smcginnis> We're still wrapping up stein for a few more weeks yet. We will be looking at new drivers once we switch focus to train.
16:25:09 <smcginnis> #topic Stable release update
16:25:14 <smcginnis> rosmaita: OK, on to you.
16:25:19 <rosmaita> the last stable releases were mid-February
16:25:27 <rosmaita> there hasn't been a lot of activity (except for cinder)
16:25:35 <rosmaita> #link https://etherpad.openstack.org/p/cinder-releases-tracking
16:25:45 <rosmaita> my idea is to put up release patches next week when things should be kind of quiet after RC-1 and before final RCs
16:25:55 <rosmaita> #link https://releases.openstack.org/stein/schedule.html
16:26:05 <rosmaita> one thing to keep in mind is that stable/pike is scheduled to go to 'extended maintenance' mode when stein is released
16:26:12 <smcginnis> rosmaita: That seems like a good point to do it.
16:26:17 <rosmaita> ok, cool
16:26:20 <rosmaita> that means we can continue to patch it, but no more releases
16:26:31 <rosmaita> there's only 1 change in os-brick stable/pike, and 1 patch with a -2
16:26:41 <rosmaita> there are 2 changes in cinder stable/pike, and 1 patch having trouble getting through the gate
16:26:45 <smcginnis> I've been meaning to send something out about extended maintenance to make sure teams are ready for that.
16:26:59 <rosmaita> someone sent something out
16:27:07 <whoami-rajat> rosmaita: are these the latest stats, i think many rocky backports merged recently.
16:27:39 <rosmaita> yes, just looked, there are 10 backports unreleased in rocky
16:27:44 <smcginnis> There are several rocky patches queued up if they can get past job failure issues.
16:27:57 <rosmaita> for pike, i could hold off on releasing those until just before stein release (week of apr 8)
16:28:07 <rosmaita> or could just release next week with the rest, and then do one more if anything new is backported
16:28:09 <whoami-rajat> smcginnis: oh yes, they didn't merge due to the tempest issue
16:28:30 <smcginnis> I think a few made it through, but a bunch still pending.
16:28:58 <rosmaita> well, i could hold off on cinder stable/rocky, or just do another one after all that stuff merges
16:29:16 <whoami-rajat> rosmaita:  smcginnis can we wait until the rocky backports make their way until pike? i saw few important ones.
16:29:37 <smcginnis> Probably no reason to hold off. They can always be picked up on the next stable release.
16:29:46 <rosmaita> whoami-rajat: so we will have to move fast on stuff going into pike
16:29:49 <smcginnis> whoami-rajat: That would be good to backport as far as we can.
16:30:12 <smcginnis> We haven't always been good about that unless someone had a specific fix they wanted to get backported.
16:30:22 <smcginnis> It would be nice to have a better process around that.
16:30:57 <rosmaita> the other thing to keep in mind, is that pike is still "open" ... just won't be able to release from it after stein release april 10 or thereabouts
16:30:58 <whoami-rajat> smcginnis: i can start cherrrypicking for previous stable branches if its ok?
16:31:17 <smcginnis> whoami-rajat: Definitely for anything that has merged in a later branch.
16:31:37 <whoami-rajat> smcginnis: ok thanks...
16:31:47 <smcginnis> rosmaita: It's not a hard and fast rule on the transition. I think if we need a few extra days to finalize things we can get it.
16:31:49 <rosmaita> smcginnis: this was the email about pike extended maintenance: http://lists.openstack.org/pipermail/openstack-discuss/2019-March/003745.html
16:32:11 <rosmaita> ok
16:32:29 <smcginnis> rosmaita: Oh yeah, I was part of the conversation that led to him sending that. :)
16:33:09 <rosmaita> ok, so sounds like stable releases next week ... what about pike? release next week and then again if necessary?
16:33:44 <smcginnis> That sounds good. Let's get out what we have, then see if there's anything else we can get backported in time to get a good final release out before EM.
16:34:02 <rosmaita> ok, great.  that's all from me
16:34:13 <smcginnis> Thanks for tracking that rosmaita
16:34:19 <rosmaita> np
16:34:29 <smcginnis> #topic Issue deploying cinder with gunicorn
16:34:39 <smcginnis> whoami-rajat: What? :)
16:34:53 <whoami-rajat> oh
16:35:12 <smcginnis> Where is gunicorn used and why do we care are my first thoughts. :]
16:35:35 <rosmaita> you are reading my mind
16:35:43 <whoami-rajat> adriant tried deploying cinder with gunicorn but faced an issue,
16:36:17 <e0ne> smcginnis: +1
16:36:34 <whoami-rajat> https://github.com/openstack/cinder/blob/master/cinder/wsgi/wsgi.py#L40
16:36:40 <e0ne> why do we need gunicorn for api?
16:36:46 <smcginnis> whoami-rajat: Is there more context to that? Why do we care?
16:37:04 <whoami-rajat> the gunicorn application func required 2 mandatory fields which wasn't present here,
16:37:27 <whoami-rajat> oh just wanted to ask if anyone tried that way, it should be possible to use other servers to deploy cinder right?
16:38:13 <smcginnis> Officially we just support running behind a web server like apache or nginx I think.
16:38:43 <e0ne> I would like to say that we support only wsgi servers for API
16:39:41 <smcginnis> Looks like gunicorn is a wsgi server
16:39:49 <smcginnis> #link https://gunicorn.org/
16:39:55 <whoami-rajat> yes it is
16:40:24 <whoami-rajat> we currently run on uWSGI IMO
16:40:33 <smcginnis> I guess if there's a non-intrusive way that we can make it not blow up with it, but doesn't have an impact on anything else, I'd be fine merging any fix for it.
16:40:51 <eharney> presumably there would be similar changes in other openstack projects?
16:41:02 <smcginnis> But I don't think we should put much time into caring about it unless there's some overall effort to make gunicorn a "standard" deployment method.
16:41:48 <whoami-rajat> eharney: adriant said he would be trying deploying keystone next and check if he face the same error, will follow up with him on that
16:42:06 <rosmaita> is gunicorn considered a production server?
16:42:20 <whoami-rajat> smcginnis: agreed, just felt like deployers should have flexibility.
16:42:48 <whoami-rajat> rosmaita: yes.
16:42:50 <smcginnis> whoami-rajat: Would be good it it was a cross-project effort.
16:43:46 <whoami-rajat> smcginnis: yes it would be, i will follow up with him regarding other projects too.
16:43:54 <e0ne> smcginnis: I agree. if only cinder supports gunicorn, it makes less sense to me even for stand alone mode
16:44:18 <whoami-rajat> oh keystone is using flask, he said he would be trying glance. my bad.
16:44:23 <e0ne> if we decide to add gunicorn support, we need to have it on CI or it will be broken
16:44:32 <smcginnis> We have so many things that need attention right now, so I'd rather we don't put much effort into something like this until it becomes more of a priority for the community.
16:44:43 <e0ne> smcginnis: +1
16:44:47 <rosmaita> agree
16:44:49 <whoami-rajat> e0ne: smcginnis  ++
16:45:05 <smcginnis> #topic Open floor
16:45:27 <smcginnis> Along those lines, another plea for folks to take a look at what we have out there and identify the things that are important to get in for stein.
16:45:31 <smcginnis> Time is running out.
16:46:40 <whoami-rajat> https://review.openstack.org/#/c/644818/ i would like to request reviews on this since this maybe the last one of api-ref samples to go in stein.
16:47:25 <smcginnis> After the RC tomorrow, we really should only get bugfixes in for any further RCs. Anything that is not a straight bug fix should get attention over the next 24 hours or so if we want it in stein.
16:47:35 <eharney> what does that patch do exactly?  says it adds "tests"?
16:49:08 <eharney> i guess i don't know how functional tests related to api-ref samples, which i thought were for documentation
16:49:09 <smcginnis> I'm wondering now if we should have written up a spec to at least document what that is all doing.
16:49:26 <smcginnis> whoami-rajat: Do you have a quick overview of how this all ties in?
16:49:29 <eharney> how functional tests are related to*
16:50:03 <smcginnis> My understanding is, the functional tests capture the API request/response that can then be used for the examples in the api-ref.
16:50:09 <e0ne> if it's only tests, why don't hold on it until Train?
16:50:16 <smcginnis> But I'm not sure what the whole process looks like.
16:50:33 <smcginnis> Yeah, also not sure it's that big of a priority that we can't just wait until train.
16:50:41 <eharney> is it only tests?  i thought api-ref was used for doc
16:50:43 <whoami-rajat> eharney:  smcginnis with new features being added, our api-ref isn't always updated, running these tests will auto generate the responses that are examples in the api-ref guide.
16:51:10 <smcginnis> whoami-rajat: Is it a manual process to then have someone take the output from the functional tests and update the api-ref?
16:51:35 <whoami-rajat> eharney: these are tests as it fails the gate when someone tries to manually interfere with the natural responses of api
16:51:40 <whoami-rajat> smcginnis: these are autogenerated
16:52:13 <smcginnis> How is the api-ref updated automatically? That's a separate job that's run in gate.
16:52:38 <whoami-rajat> on second thoughts we can hold it since bug fixes are more critical than this one. e0ne
16:53:16 <smcginnis> It is trivial enough, since it's just tests and documentation, that we could maybe still backport it later if we decide to.
16:53:24 <eharney> does stein currently have incorrect api-ref materials?  or just that some are missing?
16:53:28 <whoami-rajat> smcginnis: since i'm implementing the base responses currently its not an issue, but for every MV update there will be a separate response,
16:54:16 <smcginnis> whoami-rajat: I'm wondering if it might be useful to post something to the ML explaining what the overall process of this is so we at least have something to refer to if someone needs to understand how this is supposed to work.
16:54:17 <whoami-rajat> eharney: there were several mistakes and people were manually updating it with several patches, so gmann  proposed to implement it just like nova did to not have human errors in it.
16:54:30 <smcginnis> Or maybe better, add something to the cinder contributor docs explaining it.
16:54:54 <smcginnis> Oh right, this is following something that Nova already did, right?
16:55:07 <whoami-rajat> smcginnis: sure, i can put up a doc for it.
16:55:19 <whoami-rajat> smcginnis: yes but they haven't clearly documented it. :(
16:55:31 <smcginnis> Thanks, I think that would help having something written down.
16:55:52 <smcginnis> 4 minutes left. Anything else important to cover in the meeting?
16:56:19 <eharney> is cinderclient done for stein now?
16:56:23 <smcginnis> One final plea for folks to look for things that we should make sure we get into stein.
16:56:33 <smcginnis> eharney: Yeah, client lib freeze was a couple weeks ago.
16:56:42 <eharney> did we release it?
16:56:51 <rosmaita> 4.1.0
16:57:07 <eharney> 4.1.0 was in october..?
16:57:30 <smcginnis> *gulp*
16:57:32 <rosmaita> well, maybe not, then
16:58:12 <smcginnis> pewp - https://github.com/openstack/python-cinderclient/compare/4.1.0...master
16:58:55 <smcginnis> Nothing super critical at least. No new features.
16:59:13 <smcginnis> We can get what we really need backported and a stable release done once freeze is over.
16:59:15 <eharney> yeah, nothing huge in there at least, but should we look for a 4.1.1 soonish, or what?
16:59:45 <eharney> yeah
16:59:59 <smcginnis> OK, out of time.
17:00:01 <smcginnis> Thanks everyone.
17:00:12 <smcginnis> #endmeeting