16:00:02 <smcginnis> #startmeeting Cinder
16:00:02 <openstack> Meeting started Wed Apr 12 16:00:02 2017 UTC and is due to finish in 60 minutes.  The chair is smcginnis. Information about MeetBot at http://wiki.debian.org/MeetBot.
16:00:03 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:00:05 <openstack> The meeting name has been set to 'cinder'
16:00:13 <smcginnis> ping dulek duncant eharney geguileo winston-d e0ne jungleboyj jgriffith thingee smcginnis hemna xyang1 tbarron scottda erlon rhedlind jbernard _alastor_ bluex karthikp_ 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 watanabe.isao,tommylikehu mdovgal ildikov
16:00:18 <mdovgal> hi
16:00:19 <smcginnis> wxy viks ketonne abishop sivn breitz
16:00:21 <rajinir> o/
16:00:24 <patrickeast> Hi
16:00:25 <geguileo> Hi!
16:00:25 <tbarron> hi
16:00:26 <tommylikehu> hi
16:00:27 <jungleboyj> o/
16:00:30 <Swanson> hello
16:00:42 <e0ne> hi
16:00:45 <wxy|> hi
16:00:45 <xyang1> Hi
16:00:46 <smcginnis> Hey everyone.
16:01:12 <smcginnis> #topic Announcements
16:01:38 <smcginnis> #info Pike-1 milestone cut
16:01:45 <Rockyg> hey sean!  now I'm leaving again
16:01:53 <smcginnis> Rockyg: :)
16:02:14 <smcginnis> Since Pike-1 doesn't have any specific Cinder items, I've already done the release for that milestone.
16:02:37 <diablo_rojo_phon> Hello :)
16:02:54 <smcginnis> I think some good progress so far, but quite a few decent patches sitting out there.
16:03:04 <smcginnis> Please, please, do reviews if you have the cycles.
16:03:40 <cuhler> 0/
16:04:07 <smcginnis> And especially cores, please also review cinder-specs too. There's a few of those that we want to get in Pike that could use some attention.
16:04:30 <tommylikehu> +1
16:04:51 <smcginnis> #topic How should we clean up the expired user messages?
16:05:01 <smcginnis> tommylikehu: All yours.
16:05:09 <tommylikehu> yes, I think this part is left by sheer. As we already has this feature merged, we should consider how to clean the expired messages out.
16:05:19 <abishop> o/
16:05:30 <eharney> i think the original plan was to use a periodic task, which still seems like a good idea
16:05:31 <e0ne> IMO, cinder-manage command will be good for this
16:05:41 <smcginnis> eharney: That's what I thought.
16:05:42 <eharney> e0ne: i disagree
16:05:45 <tommylikehu> wow
16:05:47 <cFouts> o/
16:05:59 <tommylikehu> we should have a decision
16:06:00 <eharney> no real reason to push this off to a manual admin action when we can just do it from within the service
16:06:02 <smcginnis> I would like it to be automatic and not something an operator would have to remember to do.
16:06:08 <smcginnis> eharney: +1
16:06:18 <tommylikehu> I like the automatically way
16:06:22 <mdovgal> i think we can make both variants
16:06:23 <e0ne> eharney: good point
16:06:33 <tommylikehu> mdovgal why both
16:06:42 <jungleboyj> smcginnis: ++
16:06:46 <smcginnis> mdovgal: We could, but is there a benefit?
16:07:18 <e0ne> I won't argue if community decides to have periodic job for it
16:07:42 <mdovgal> tommylikehu, why not? maybe i need to clean up it immediately, not to wait when periodic task will be triggered
16:07:51 <smcginnis> Anyone have any argument for not making it a periodic task that's done automatically?
16:08:08 <tommylikehu> smcginnis:  we can delete when inserting
16:08:17 <tommylikehu> once a day maybe
16:08:40 <mdovgal> to which service will this task be connected?
16:08:45 <tommylikehu> but I prefer the perodic task
16:08:54 <eharney> cinder-volume i'd assume
16:09:04 <smcginnis> mdovgal: I think you are right we could do both. But I think I would rather make it automatic first, then wait to see if there is operator feedback that they need a manual method before we implement that.
16:09:09 <smcginnis> mdovgal: That make sense?
16:09:26 <mdovgal> smcginnis, i think it's ok
16:09:32 <cFouts> smcginnis: +1
16:09:32 <smcginnis> mdovgal: Cool
16:09:42 <geguileo> How are we going to trigger the automatic removal?
16:09:59 <tommylikehu> when expired
16:10:13 <geguileo> No, I mean how the task is going to be fired
16:10:19 <tommylikehu> check the records after a period
16:11:09 <mdovgal> tommylikehu, just to check count of records? or datetime of them?
16:11:30 <tommylikehu> I think the datetime would be ok
16:11:43 <tommylikehu> or the expired_at column
16:11:53 <geguileo> tommylikehu: How is the method that does the cleanup going to be started/run?
16:12:07 <cFouts> geguileo: periodic_task?
16:12:24 <jungleboyj> geguileo:  Yeah, periodic task in cinder-volume?
16:12:27 <geguileo> cFouts: where? and how do you make sure it only runs once and not in multiple services?
16:12:36 <cFouts> ah
16:12:41 <jungleboyj> Oh ...
16:12:42 <tommylikehu> why not scheduler
16:12:45 <geguileo> jungleboyj: What happens when you have multiple c-vol services?
16:12:49 <geguileo> tommylikehu: Same issue
16:12:54 <tommylikehu> ok
16:12:57 <jungleboyj> geguileo:  Yeah, that just hit me.
16:12:59 <geguileo> We don't have a leader, so they could all trigger it
16:13:05 <smcginnis> geguileo: Good point. But would that really cause a problem if multiple services ran it?
16:13:25 <geguileo> smcginnis: Depends on the queries they do and how many services you have running them
16:13:30 <geguileo> as they could come in waves
16:14:00 <geguileo> evey though you could add a randomization to the period of the start of the task
16:14:19 <smcginnis> geguileo: Dont' know, but wouldn't it just be a DELETE WHERE expired_at > now, so one service would delete all records, the rest would just pass.
16:14:39 <geguileo> smcginnis: If that's all it really takes then it's ok
16:14:47 <smcginnis> Yeah, not sure.
16:15:00 <tommylikehu> smcginnis: +1
16:15:02 <geguileo> just wanted to bring it up in case it turned out to be more complex
16:15:06 <smcginnis> tommylikehu: But I think we have an answer. We can debate implementation details in a review if you want to implement something.
16:15:16 <wxy|> FYI, this's what glance does for the expired task: when users list tasks, glance will delete the expired ones at the same time.
16:15:32 <smcginnis> wxy|: Only on the list request?
16:15:39 <wxy|> smcginnis: yeah
16:16:00 <smcginnis> Problem I see with that is if no one asks for them, they just accumulate.
16:16:08 <smcginnis> Not sure if that would actually be the case though.
16:16:09 <tommylikehu> it could
16:16:19 <smcginnis> But I kind of like the periodic task idea better at this point.
16:16:31 <tommylikehu> smcginnis: I can pick up that implementation
16:16:45 <smcginnis> tommylikehu: Great, thanks! OK if we move on then.
16:16:51 <tommylikehu> thanks
16:16:53 <wxy|> https://github.com/openstack/glance/blob/master/glance/db/sqlalchemy/api.py#L1502
16:17:04 <smcginnis> wxy|: Thanks
16:17:07 <e0ne> cinder-manage with cron job could be an option in a future
16:17:09 <smcginnis> #topic How should we support like filters for name/description etc columns?
16:17:32 <smcginnis> e0ne: Could work too, but requires outside configuration. Certainly a possibility though.
16:17:43 <tommylikehu> let's assume we would have like filter in the future :)
16:17:53 <smcginnis> :)
16:18:19 <tommylikehu> will we support both exact-match and inexact-match for one columns?
16:18:36 <smcginnis> jgriffith: You around? I know you had some opinions on filtering.
16:18:46 <jgriffith> I am now :)
16:18:47 <tommylikehu> something like  'cinder list --filters name=vol' and 'cinder list --filters name%=vol'
16:18:50 <jgriffith> Literally just logged on
16:19:07 <tommylikehu> jgriffith:  thanks
16:19:17 <smcginnis> jgriffith: Timing is everything. ;)
16:19:48 <jgriffith> so I'm not a fan of the "filter on all things", that being said I talked to tommylikehu last night and if people would like this functionality that's cool; we could work it into the proposed generic filter design that I proposed
16:20:22 <jgriffith> but honestly a lot of this is overkill IMO, if people need this that badly just write your own client/shell script to do it
16:20:32 <jgriffith> that's all I have :P
16:21:25 <tbarron> +1 for integrating it into the proposed generic filter design
16:21:32 <smcginnis> tbarron: Agree
16:21:54 <tommylikehu> jgriffith, tbarron  I just updated my spec to take advantages of jgriffith 's pne
16:21:59 <tommylikehu> one
16:22:00 <tommylikehu> https://review.openstack.org/#/c/442982/
16:22:52 <tommylikehu> it seems that no one cares about this topic on whether we should support them bothv:(
16:22:58 <tommylikehu> both :(
16:23:01 <smcginnis> I do think if we're going to provide filtering on the service side, it would be a better user experience to be able to do both.
16:23:35 <tommylikehu> yes :)
16:23:38 <tbarron> smcginnis: +1
16:24:02 <smcginnis> I think folks care but are still thinking through it. ;)
16:24:27 <smcginnis> tommylikehu: OK, unless anyone brings up a strong argument against both, I think we have an answer.
16:24:44 <tommylikehu> thanks, that's clear
16:24:54 <smcginnis> #topic remotefs based 3rd party CIs broken
16:25:08 <smcginnis> This was added by kaisers1 but he wasn't able to attend
16:25:27 <smcginnis> remotefs drivers are broken right now apparently.
16:25:35 <smcginnis> https://bugs.launchpad.net/cinder/+bug/1681406
16:25:36 <openstack> Launchpad bug 1681406 in Cinder "RemoteFS: volume cascade delete fails" [Undecided,In progress] - Assigned to Lucian Petrut (petrutlucian94)
16:25:53 <smcginnis> #link  https://review.openstack.org/#/c/455256/ Proposed fix.
16:26:14 <smcginnis> So I think he wanted this in the meeting agenda to just raise awareness.
16:26:23 <smcginnis> Consider it raised.
16:26:34 <jungleboyj> I feel aware.
16:26:53 * smcginnis imagines jungleboyj glowing with enlightenment
16:27:00 <smcginnis> :)
16:27:07 <smcginnis> #topic Open Discussion
16:27:10 * jungleboyj laughs ... If only.
16:27:13 <smcginnis> Anything else to discuss today?
16:27:18 <smcginnis> Summit is coming up.
16:27:23 <smcginnis> Hope to see a lot of folks there.
16:27:35 <smcginnis> Or at least at "the forum"
16:27:42 <jungleboyj> Looked like hotels were filling up so ...
16:27:42 <mdovgal> I have one little question
16:28:45 <Swanson> ?
16:28:55 <mdovgal> there is one blueprint about cascade backup delete https://blueprints.launchpad.net/cinder/+spec/cascade-delete-backup
16:29:01 <mdovgal> #link https://blueprints.launchpad.net/cinder/+spec/cascade-delete-backup
16:29:29 <mdovgal> Do somebody have any problems with it implementation in future?
16:29:30 <e0ne> it should be a useful feature for incremental backups
16:29:58 <xyang2> might be nice to have this
16:30:12 <smcginnis> mdovgal: Just approved it.
16:30:19 <cuhler> Another question: We're working on CI for Veritas os-brick connector, https://review.openstack.org/#/c/442754. Is CI needed before merge?
16:30:24 <cuhler> Nova team said they want connector merged before starting to review the Veritas Nova volume driver that depends on it (https://review.openstack.org/#/c/443951)
16:30:29 <mdovgal> smcginnis, nice. thank you
16:30:36 <smcginnis> mdovgal: No problem.
16:30:45 <diablo_rojo> cuhler, its not required no, but its strongly advised
16:30:59 <smcginnis> cuhler: Our policy has been if you have a specific connector in os-brick then it is strongly advised to have CI coverage.
16:31:19 <smcginnis> cuhler: I mainly wanted some record on there as to the plans for that, and I see it was added.
16:31:30 <smcginnis> cuhler: I don't think we need to block the review waiting on that.
16:31:41 <cuhler> Okay. Can the CI be driven by the unmerged Nova volume driver, or is Nova somehow simulated?
16:32:05 <smcginnis> cuhler: Yeah, I think it would have to be a cherry pick of your driver into nova or something.
16:32:16 <smcginnis> cuhler: Until that merges in Nova at least.
16:32:21 <cuhler> ok, that clarifies things. Thx
16:32:33 <cuhler> done
16:33:59 <smcginnis> Oh, next PTG announced in case anyone missed that.
16:34:12 <smcginnis> Denver the week of Sept 11.
16:34:28 <diablo_rojo> MOAR torchy's tacos!
16:34:34 <smcginnis> :)
16:34:36 <Swanson> Denver? Spit.
16:34:42 <jungleboyj> :-)  I had that thought too.
16:34:45 <smcginnis> Bugsmash even taking place the week after the summit.
16:34:53 <smcginnis> I like Denver!
16:35:02 <Swanson> You would.
16:35:08 <jungleboyj> smcginnis: ++
16:35:08 <smcginnis> I've heard their brownies are fantastic.
16:35:09 <xyang2> cool!  no international travel
16:35:11 <smcginnis> :D
16:35:18 <mdovgal> Not NBA time on the 11 of Sept :(
16:35:25 <diablo_rojo> It was cheaper than our other option and ATL both so hopefully that helps people get there.
16:35:33 <jungleboyj> Sounds like the PTG after that will international though.
16:35:33 <smcginnis> +1
16:35:55 <xyang2> jungleboyj: is it?
16:35:57 <smcginnis> Actually glad to see the PTG alternate since we were never able to get that going with the midcycle.
16:36:09 <jungleboyj> smcginnis: True.
16:36:30 <jungleboyj> xyang2:  Yeah, I think they were talking Europe.
16:36:37 <jungleboyj> diablo_rojo:  Probably knows more.  :-)
16:36:49 <xyang2> jungleboyj: well, too early to worry about that now:)
16:37:07 <smcginnis> :)
16:37:15 <smcginnis> Anything else today?
16:37:24 <diablo_rojo> jungleboyj, I know of places we are looking at but we are FAR from a decision so I don't know all that much more :)
16:37:37 <jungleboyj> diablo_rojo:  Cool.
16:37:54 <smcginnis> diablo_rojo: Then next two summits are Sydney and Vancouver, right.
16:38:07 <diablo_rojo> smcginnis, yep :)
16:38:44 <smcginnis> OK, sounds like we are pretty much done here. Thanks everyone.
16:38:50 <smcginnis> Please do reviews! :)
16:38:50 <Swanson> United tickets should be cheap and plentiful. Buy your airfare now.
16:38:55 <smcginnis> LOL
16:38:59 <e0ne> :)
16:39:02 <viks> :D
16:39:10 <jungleboyj> Bwah ha ha
16:39:10 <xyang2> Swanson: Never United!:)
16:39:13 <smcginnis> Swanson: At this rate you can just buy the whole plane.
16:39:21 <viks> they are not that cheap still
16:39:27 <jungleboyj> smcginnis:  That would be awesome!
16:39:39 <smcginnis> #endmeeting