16:00:45 <smcginnis> #startmeeting Cinder
16:00:45 <openstack> Meeting started Wed Jan 18 16:00:45 2017 UTC and is due to finish in 60 minutes.  The chair is smcginnis. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:46 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:00:49 <openstack> The meeting name has been set to 'cinder'
16:00:53 <bswartz> .o/
16:00:53 <smcginnis> ping dulek duncant eharney geguileo winston-d e0ne jungleboyj jgriffith thingee smcginnis hemna xyang1 tbarron scottda erlon rhedlind jbernard _alastor_ bluex patrickeast dongwenjuan JaniceLee cFouts Thelo vivekd adrianofr mtanino yuriy_n17 karlamrhein diablo_rojo jay.xu jgregor baumann rajinir wilson-l reduxio wanghao thrawn01 chris_morrell stevemar watanabe.isao,tommylikehu mdovgal ildikov
16:00:54 <mdovgal> hi all
16:00:55 <_alastor_> o/
16:01:00 <smcginnis> wxy viks ketonne
16:01:01 <smcginnis> Hey everyone
16:01:01 <kaisers> o/
16:01:02 <geguileo> Hi o/
16:01:02 <rarora> hi
16:01:02 <diablo_rojo> Hello :)
16:01:03 <hemna> yough
16:01:04 <scottda> hi
16:01:10 <viks> hi
16:01:12 <erlon> hey
16:01:15 <eharney> hi
16:01:15 <stevemar> o/
16:01:16 <rajinir> hi
16:01:20 <jungleboyj> Hello.
16:01:27 <patrickeast> Hi
16:01:33 <smcginnis> stevemar: Yay, we're turning you into a Cinder person. :P
16:01:34 <xyang1> hi
16:01:34 <erlon> smcginnis: wxy viks ketonne!
16:01:34 <wxy|> hi
16:01:35 <markstur> hi
16:01:39 <e0ne> hi
16:01:54 <DuncanT> Hi
16:01:58 <jungleboyj> stevemar: Yay!
16:02:03 <stevemar> smcginnis: jungleboyj o_O
16:02:07 <smcginnis> hehe
16:02:13 * stevemar leaves quickly before he turns into a brick
16:02:23 <e0ne> :)
16:02:31 <jungleboyj> stevebaker: Is that a drunk look?  If so, you fit right in.
16:02:32 <smcginnis> #topic Announcements
16:02:36 <dulek> o/
16:02:43 <smcginnis> #link https://etherpad.openstack.org/p/cinder-spec-review-tracking Review focus
16:03:04 <smcginnis> #info Sprint to focus on HA and attach/detach today and tomorrow.
16:03:16 <smcginnis> That etherpad has the outstanding patches for both of these efforts.
16:03:40 <smcginnis> If we an get some focus on reviewing and testing those, it would be awesome if we can knock those out.
16:03:42 <Swanson> hello
16:04:02 <erlon> smcginnis: when is the ff?
16:04:11 <smcginnis> It's getting close to the end of the cycle, so it always gets me nervous pushing things like that through too close to have time to fix anything that comes up later.
16:04:43 <smcginnis> erlon: Feature freeze is next week already.
16:04:51 <smcginnis> #link https://releases.openstack.org/ocata/schedule.html Ocata schedule
16:04:56 <erlon> smcginnis: Tuesday?
16:05:00 <smcginnis> So we really are getting down to the wire here.
16:05:29 <smcginnis> erlon: Thursday's are the offical days, but I like to get things done a day or two early whenever possible just in case.
16:05:46 <erlon> smcginnis: will it be possible to get some focus in the BFS Snapshots feature?
16:05:58 <smcginnis> NFS I'm assuming. :)
16:06:13 <smcginnis> erlon: Yeah, that would really be another good one to get in there.
16:06:16 <erlon> smcginnis: I have some people on my company counting on that to be in Ocata
16:06:27 <erlon> smcginnis: :)
16:06:28 <jungleboyj> smcginnis: erlon +2
16:06:41 <smcginnis> erlon: What's the status of that now? Everything testing out right?
16:06:44 <hemna> erlon, is that review in the list?
16:06:49 <hemna> the ether pad?
16:07:00 <erlon> hemna: yes
16:07:02 <smcginnis> hemna: Farther down.
16:07:42 <smcginnis> #link http://www.openstack.org/ptg PTG info and registration
16:07:53 <smcginnis> Please make every effort to attend the PTG.
16:08:12 <smcginnis> We will hopefully get back to streaming on YouTube  if we can for those that can't attend in person.
16:08:17 <erlon> hemna: smcginnis: jungleboyj: righ now there something failing, but I would ask to people still review it, as that shouldn't be hard to solve. Ill get to that ASAP
16:08:25 <hemna> erlon, https://review.openstack.org/#/c/147186/  that one?
16:08:26 <smcginnis> Just need a voice tracking drone to fly around the room and record us.
16:08:34 <erlon> hemna: yes
16:08:37 <smcginnis> erlon: ack
16:08:37 <hemna> kew
16:08:48 <jungleboyj> He he ... That would be awesome.
16:09:08 <smcginnis> diablo_rojo: I think you said Cinder representation is looking OK with the latest registration counts?
16:09:25 <erlon> smcginnis: will hemma attend? I think he is the drone'a owner
16:09:37 <smcginnis> :)
16:09:48 <smcginnis> hemna: You are able to go this time, right?
16:09:55 <hemna> yah I got approval
16:10:00 <smcginnis> Woot
16:10:03 <hemna> I just have to book my flight and hotel still
16:10:07 <tbarron> hi
16:10:10 <e0ne> good news!
16:10:13 <stevemar> there are 90 tickets left :)
16:10:24 <smcginnis> stevemar: Oh, not bad.
16:10:25 <stevemar> it was 130 about 10 days ago
16:10:32 <smcginnis> Rush at the end.
16:10:34 <stevemar> it was at 130 for a looong time
16:10:37 <stevemar> yep
16:10:55 <stevemar> i think folks were just waiting for the new year to get approval (at least I was)
16:10:57 <smcginnis> Hard part I think is a lot of companies won't allow booking everything until it falls under the right fiscal quarter.
16:11:01 <ildikov> stevemar: management ususally does not like early decisions :)
16:11:03 <jungleboyj> smcginnis: I got official approval as well.  Just working out details.
16:11:12 <diablo_rojo> hemna, You should use the offical hotels if you can :) It would really help us out. We need to hit our hotel numbers.
16:11:13 <smcginnis> jungleboyj: Awesome!
16:11:31 <diablo_rojo> smcginnis, +1
16:11:33 <ildikov> stevemar: at least that's my experience re late approvals
16:11:47 <diablo_rojo> ildikov, sad but true
16:11:47 <jungleboyj> diablo_rojo: What happens if you miss those numbers?
16:12:01 <smcginnis> It comes out of diablo_rojo's paycheck. :)
16:12:03 <diablo_rojo> jungleboyj, The foundation has to pay a lot of extra money for the venue.
16:12:32 <jungleboyj> diablo_rojo: Ok.  The conference rate isn't good and companies want to use the Hilton instead.
16:12:40 <diablo_rojo> jungleboyj, to get the number of conference rooms we have we had to lock in a bunch of hotel rooms and so if no one reserves them we have to pay for them.
16:12:43 <smcginnis> diablo_rojo: Any idea how short we are?
16:12:50 <diablo_rojo> smcginnis, a lot :/
16:12:59 <diablo_rojo> but the more we get the less it will hurt
16:13:09 <smcginnis> diablo_rojo: That's why I've never liked that model. :/
16:13:16 <smcginnis> OK...
16:13:17 <smcginnis> #link https://www.openstack.org/summit/boston-2017/call-for-presentations/ Summit CFP
16:13:18 <jungleboyj> diablo_rojo: Ok, I will see what I can do.
16:13:19 <diablo_rojo> jungleboyj, Yeah I've heard that from a few people.
16:13:30 <dulek> Well, it kind of doesn't make sense that PTG was supposed to be cheap, and then Foundation requires us to book a more expensive hotel.
16:13:32 <diablo_rojo> jungleboyj, thanks :)
16:13:40 <jungleboyj> diablo_rojo: Yep.  :-)
16:13:59 <e0ne> dulek: +1
16:13:59 <jungleboyj> dulek: So, is it really cheap?
16:14:05 <hemna> dulek, +1
16:14:13 <smcginnis> Session proposals for the Summit are now open.
16:14:16 <hemna> the PTG and the summit changes have been a disaster since the start
16:14:18 <hemna> meh
16:14:25 <jungleboyj> smcginnis: What is the deadline for those?
16:14:29 <diablo_rojo> dulek, The type of event it is is kind of a special unicorn in regards to the type of space we require and so that throws a monkey wrench into things.
16:14:32 <jungleboyj> hemna: +2
16:14:35 <smcginnis> Given the switch from the Design Summit, might be a good way to get extra management incentive to attend if you can get a talk accepted.
16:14:40 <smcginnis> jungleboyj: Good question...
16:14:41 <bswartz> hemna: pessimistic much?
16:14:44 <diablo_rojo> hemna, growing pains :)
16:14:53 <Swanson> "special unicorn" vs those "ordinary unicorns".
16:14:54 <ildikov> jungleboyj: Febr 6 IIRC
16:14:56 <hemna> bswartz, always :)
16:15:00 <smcginnis> ildikov: Beat me to it!
16:15:06 <scottda> Please make all comlaints in #openstack-B&*$tch
16:15:08 <ildikov> smcginnis: :)
16:15:12 <dulek> diablo_rojo: I'm not saying that it's easy, I understand the difficulties.
16:15:21 <smcginnis> bswartz: You must not know hemna well if you have to ask that. :D
16:15:28 <hemna> smcginnis, :)
16:15:29 <jungleboyj> ildikov: Thank you!
16:15:34 <diablo_rojo> Swanson, :P
16:15:54 <smcginnis> There is an #openstack-ptg unofficial channel for any PTG related stuff.
16:16:13 <smcginnis> #link https://etherpad.openstack.org/p/ATL-cinder-ptg-planning PTG topic planning
16:16:25 <xyang1> smcginnis: how come that is unofficial
16:16:32 <smcginnis> Please add any topics to the etherpad that you think would be good to discuss.
16:16:51 <smcginnis> xyang1: Just not sure if it's logged and stuff. Though it might be now, not really sure.
16:17:03 <xyang1> ok
16:17:12 <dulek> smcginnis: Only VIPs in there right now. ;)
16:17:22 <smcginnis> So far the PTG is going to be scottda talking for three days, so please add any extra topics. :)
16:17:25 <smcginnis> dulek: Hah!
16:17:43 * scottda looks away
16:17:44 * dulek likes scottda's talking, what's the problem?
16:17:51 <bswartz> scottda talking for 3 days works for me
16:17:54 <smcginnis> Works for me. :)
16:18:03 <diablo_rojo> scottda, I will play your favorite song so you can have a dance break from talking ;)
16:18:06 <_alastor_> it seems like if the PTG can't go on without a certain number of hotel reservations, it should just include the price of the hotel in the ticket
16:18:12 <bswartz> as long as jgriffith writes on the whiteboard at the same time
16:18:15 * jungleboyj is laughing
16:18:24 <smcginnis> bswartz: LOL
16:18:24 <scottda> but he has to write LARGE
16:18:38 <smcginnis> #link http://lists.openstack.org/pipermail/openstack-dev/2017-January/109899.html Migrate CGs to Generic Volume Groups update
16:18:48 <jungleboyj> Caddilac Caddilac
16:18:51 <smcginnis> Reminder about CGs and volume groups
16:19:00 * diablo_rojo is going to bring good whiteboard markers
16:19:02 <smcginnis> xyang1: Want to comment on that at all?
16:19:08 <diablo_rojo> jungleboyj, Pop that trunk!
16:19:28 <hemna> no magic whiteboards?
16:19:31 <xyang1> smcginnis: I have not got acknowledgement from some yet
16:19:38 <jungleboyj> :-0
16:19:41 <jungleboyj> :-)
16:20:02 <smcginnis> Now now children... :)
16:20:03 <xyang1> scottda: hemna have IBM driver maintainers got the msg
16:20:12 <hemna> xyang1, ?
16:20:17 <hemna> "the message"?
16:20:21 <xyang1> I did send out emails to driver maintainers
16:20:23 <scottda> xyang1: just the overview
16:20:30 <hemna> ok, I didn't see it
16:20:38 <xyang1> hemna: the migrate cg to groups one
16:20:41 <diablo_rojo> smcginnis, 'Our beloved ptl and head cat herder'
16:20:44 <scottda> xyang1: We know we've work to do by Pike-1
16:20:48 <xyang1> hemna: what is your email now
16:20:49 <smcginnis> Hah
16:20:58 <Swanson> xyang1, I have that email open actually and I plan to read it or at least have smcginnis give me the gist of it.
16:21:17 <smcginnis> Swanson: :)
16:21:22 * jungleboyj is going to call smcginnis Dad now.
16:21:32 <xyang1> Swanson: ok, thanks
16:21:35 <smcginnis> jungleboyj: Go to your room.
16:21:56 <smcginnis> #topic Testing for Ocata release
16:22:09 <smcginnis> scottda: Taker er away. Please.  :]
16:22:13 <scottda> #link https://etherpad.openstack.org/p/cinder-ocata-testing
16:22:15 * jungleboyj sulks away.
16:22:56 <scottda> We discovered in Newton that our testing is ad hoc for features without tempest or functional tests.
16:23:18 <scottda> So, a bit too late, we had folks sign up for whatever they were already testing...
16:23:23 <scottda> mainly, to identify gaps.
16:23:39 <scottda> So, it'd be nice if you're testing something to sign on that etherpad.
16:23:48 <scottda> And add features that need manual testing.
16:24:08 <scottda> Mainly, we just need to make sure *someone* is testing stuff before we ship.
16:24:40 <hemna> whoa whoa whoa....before?
16:24:54 <e0ne> hemna: :)
16:24:58 <scottda> Yes, radical new idea. We've fired all the customers as beta testers.
16:25:03 <jungleboyj> hemna: I know.  Crazy, isn't it?
16:25:04 <diablo_rojo> hemna :)
16:25:22 <scottda> Anyway, that's it. Nice to test stuff.
16:25:26 <Swanson> This is against the open source credo of the customer as tester.
16:26:16 <scottda> I'm done. Please help.
16:26:40 <smcginnis> scottda: OK, thanks.
16:27:17 <smcginnis> #topic HA and Attach/Detach sprint
16:27:36 <smcginnis> OK, I already ended up covering this in the announcements, but just for good measure.
16:27:59 <smcginnis> Please use the review focus etherpad to try to get some of these through.
16:28:33 <smcginnis> I mentioned in the ML post, it would be good to get these in, then fix any bugs we fine.
16:28:44 <smcginnis> The main thing being that we do not break any existing functionality.
16:29:10 <smcginnis> We tend to get held up on things. I'd like to see some actual run time on some of this code before it's too late.
16:29:35 <geguileo> If HA A/A breaks anything, please ping me and I'll fix it asap
16:29:46 <geguileo> As we've already seen a couple of bugs crawl in there
16:29:52 <smcginnis> geguileo: +1
16:29:58 * scottda puts Gorka's number on his company's support list
16:30:03 <smcginnis> geguileo: Thanks for staying on top of all that.
16:30:09 <smcginnis> scottda: Hah, good idea. :)
16:30:16 <geguileo> lol
16:30:28 <smcginnis> That's the kind of thing I'd like to see shaken out before the final cut though.
16:30:54 <dulek> geguileo: Are we planning to mark any driver as A/A-supported?
16:30:57 <smcginnis> I would hate to merge the patches in right before the freeze and find out a few weeks into Pike that we broke something horribly.
16:31:04 <dulek> geguileo: Otherwise people won't really be able to test this.
16:31:12 <dulek> Or am I wrong somehow?
16:31:37 <geguileo> dulek: I'm more worried about people testing existing functionality with that code in
16:31:56 <scottda> geguileo: Are there  any patches in the queue that are dangerous? or safer?
16:31:57 <geguileo> dulek: Less concerned at this point with the A/A functionality being perfect
16:32:07 <smcginnis> I think Pike will be when we can really say it is supported once driver maintainers get a chance to validate that setup.
16:32:10 <dulek> geguileo: Hey, we've just talked 'bout users being beta testers. :D
16:32:35 <scottda> It seems we've already merged most of the important stuff.
16:32:38 <dulek> I wondered if we're able to get e.g. Ceph changes in Ocata.
16:32:39 <geguileo> scottda: The more dangerous one is the replication one that is the next on the review list
16:32:58 <smcginnis> Eh, no one cares about replication anyway. :P
16:33:07 <jungleboyj> scottda: Why didn't I think of that when talking about our support earlier.  Just forward all calls to geguileo.  Problem solved!
16:33:09 <geguileo> The most dangerous one was the scheduler one that already merged
16:33:25 * jungleboyj glares at smcginnis
16:33:41 <dulek> geguileo: Meh, only 2 bugs in there. ;)
16:33:47 <smcginnis> jungleboyj: Hey, did I say you could come out of your room yet?
16:33:54 <scottda> geguileo: Yeah, I was thinking that the stuff that might break things are mostly already in.
16:33:58 <dulek> We break manage-unmanage each milestone, it's normal. :D
16:34:00 <Swanson> schmeplication.
16:34:04 * jungleboyj goes back to my room and closes the door
16:34:05 <smcginnis> dulek: :)
16:34:10 <geguileo> dulek: But big ones  XD
16:34:16 <smcginnis> jungleboyj: Hehe.
16:34:25 <smcginnis> jungleboyj: You have such a long commute now.
16:34:27 <dulek> But seriously - we should test retype, migration and manage-unmanage in the gate…
16:34:39 <jungleboyj> :-)  Yep.
16:34:48 <geguileo> dulek: +1
16:34:51 <smcginnis> dulek: We do seem to break that every other month.
16:35:11 <smcginnis> #topic Open Discussion
16:35:35 <scottda> FYI, I posted this on the ML: http://lists.openstack.org/pipermail/openstack-dev/2017-January/110043.html
16:35:41 <smcginnis> Anything else to cover today?
16:35:42 <_alastor_> I'm curious how we would test manage-unmanage in gate
16:35:50 <scottda> About having consistent endpoints in the service catalog...
16:35:57 <smcginnis> Ehh, my favorite topic.
16:36:02 <jungleboyj> _alastor_: Hmmm, good question.
16:36:03 <scottda> We now have 'volume', volumev2', and 'volumev3'
16:36:06 <scottda> Which sux
16:36:06 <_alastor_> since it requries backend specific information
16:36:18 <smcginnis> _alastor_: It's in there now for LVM.
16:36:23 <smcginnis> At the expense of everyone else...
16:36:28 <Swanson> In gate wouldn't you just do one backend? CI's would be a problem.
16:36:29 <erlon> _alastor_: create a volume and them manage it?
16:36:30 <smcginnis> scottda: I agree completely.
16:36:36 <dulek> smcginnis: That's snapshots only I think.
16:36:41 <viks> hi https://review.openstack.org/#/c/417537/
16:36:43 <scottda> I just submitted this fix: https://review.openstack.org/422078
16:36:46 <eharney> yeah, i think manage/unmanage volume is missing still?
16:36:49 <smcginnis> dulek: True
16:36:58 <smcginnis> Same issue though.
16:36:59 <erlon> _alastor_: shouldnt be hard to derive it for LVM once you have the volume id
16:37:12 <dulek> _alastor_, smcginnis: We could use list manageable if it was required from drivers. Then Tempest imlementation gets easy.
16:37:22 <smcginnis> I do think we should look at catalog endpoints again.
16:37:23 <scottda> OK, when we are done with manage-unmange discussion....
16:37:24 <_alastor_> erlon: Maybe each driver should define a function that will create an "unmanaged" volume on the backend to be managed by thetest
16:37:28 <_alastor_> *test
16:37:30 <bswartz> automated tests for manage/unmanage is nontrivial work
16:37:34 <erlon> _alastor_: I mean: create a volume, get the prov_info, unmanage it, manage it
16:37:36 <smcginnis> It really should just be one "volume" endpoint, then discover version from there.
16:37:40 <smcginnis> At least IMO.
16:37:42 <bswartz> we did it in manila and we ended up redesigning the whole feature for testability
16:37:46 <scottda> smcginnis: Yes, we'll discuss this at the PTG...
16:37:50 <Swanson> No list manageable. I'm not coughing up a list of every volume to anyone on the system.
16:37:57 <jungleboyj> smcginnis: +2
16:38:07 <scottda> smcginnis: I'd like cinder folks to have a consensus on what we can and will do to change all that.
16:38:17 <smcginnis> scottda: +1
16:38:20 <erlon> _alastor_: drivers barely have functions to work properly
16:38:31 <smcginnis> Good to get folks thinking about it now ahead of the PTG so we have a chance to make a decision there.
16:38:39 <xyang1> bswartz: the redesign was before the feature was merged though
16:38:39 <erlon> _alastor_: that wouldn't something reliable
16:38:45 <bswartz> it's not ideal but you can get minimal test coverage by testing create, unmanage, manage, delete
16:38:50 <smcginnis> scottda: Although maybe we should just come up with a PoC patch to do it.
16:39:13 <scottda> smcginnis: Well, it's a keystone thing, And we also don't want to break any users.
16:39:13 <bswartz> xyang1: yes, I'm just trying to point out how hard it was and what we have to do to actually get working tests for this feature
16:39:16 <erlon> _alastor_: I think that the list_manageable volumes would solve that
16:39:20 <wxy|> smcginnis: My colleague told me that Our CI is working now. Can we now remove the unsupported tag from Huawei volume driver?
16:39:25 <scottda> smcginnis: So, we'll have to think about deprecation, etc.
16:39:26 <xyang1> bswartz: agree
16:39:27 <smcginnis> bswartz: That problem beyond LVM is the manage call may be different for each backend.
16:39:34 <erlon> _alastor_: but not every driver implements it
16:39:51 <smcginnis> wxy|: I'll look again. Logs were not accessible. Otherwise it looked like tests were passing.
16:40:05 <_alastor_> erlon: I guess (at least with my implementation) that wouldn't work because any volume created by OpenStack is already unmanageable since that function assumes it's already managed
16:40:08 <smcginnis> wxy|: If we can get access to the logs, we can revert that unsupported patch.
16:40:13 <bswartz> smcginnis: In manalia we solved this by requiring drivers to provide a string at volume creation time which is guaranteed to work for re-managing that volume if it ever gets unamanged
16:40:16 <kaisers> smcginnis: regarding similar drivers, should we ping you personally?
16:40:29 <bswartz> smcginnis: it was the only way we could ensure the feature would work in a CI environment
16:40:41 <kaisers> smcginnis: i mean for reverting the unsupported flag of drivers whose CIs are back
16:40:53 <smcginnis> kaisers: Just ping the #openstack-cinder channel.
16:40:59 <_alastor_> erlon: Which is why I was thinking a second function for normal "non-openstack" volume creation would be helpful for thetest
16:41:00 <xyang1> bswartz: actually there are still driver options in manila
16:41:01 <erlon> _alastor_: hmm, you would need to create some volumes, unmanaged then, and then you would see something listed
16:41:02 <kaisers> smcginnis: ok
16:41:11 <_alastor_> erlon: yeah, or that :)
16:41:29 <bswartz> xyang1: the test relies on that export location string being sufficient though, doesn't it?
16:42:05 <xyang1> bswartz: not enough, there is a recent patch from ganso to use driver options in ci
16:42:19 <erlon> _alastor_: internally we have a a test harness that do that, checking directly in the backend, but to specific to us
16:42:46 <bswartz> xyang1: >_<
16:43:40 <xyang1> bswartz: I am talking about manage snapshot.  manage volume might be ok
16:43:47 <bswartz> ok
16:44:12 <tbarron> can remove one topic from the manila meeting tomorrow :D
16:44:20 <xyang1> tbarron: :)
16:44:22 <bswartz> lol
16:44:41 <smcginnis> Anything else today?
16:44:58 <smcginnis> OK, thanks everyone.
16:45:01 <viks> hi https://review.openstack.org/#/c/417537/
16:45:06 <jungleboyj> Thank you!
16:45:07 <smcginnis> jungleboyj: You can come out of your room now. :P
16:45:09 <jungleboyj> Fun meeting!
16:45:21 <jungleboyj> smcginnis: Thank you.  I will be good now.
16:45:22 <scottda> good thing you remembered to set him free smcginnis
16:45:27 <diablo_rojo> smcginnis, as always, thanks for putting up with us :)
16:45:30 <scottda> Last week he was up there for 2 days.
16:45:31 <smcginnis> #endmeeting