15:00:21 <gordc> #startmeeting ceilometer
15:00:24 <openstack> Meeting started Thu Jul 30 15:00:21 2015 UTC and is due to finish in 60 minutes.  The chair is gordc. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:25 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:27 <sileht> o/
15:00:28 <openstack> The meeting name has been set to 'ceilometer'
15:00:29 <_nadya_> o/
15:00:30 <cdent> o/
15:00:42 <fabiog> hi
15:01:08 <prad> o/
15:02:10 <gordc> let's say this is enough since i see a bunch of other names present.
15:02:21 <gordc> #topic recurring: roadmap items (new/old/blockers) https://wiki.openstack.org/wiki/Ceilometer/RoadMap
15:02:36 <gordc> so we cut liberty 2 this morning
15:02:40 <gordc> https://launchpad.net/ceilometer/+milestone/liberty-2
15:03:05 <gordc> the above is what was included. had to defer a few things since gate is what it is and nothing is getting in any time soon
15:03:24 <gordc> https://blueprints.launchpad.net/ceilometer/liberty
15:03:53 <gordc> we seem pretty well on track for liberty... for the remaining open items please let someone know if there's any issues.
15:03:57 <jasonamyers> o/
15:04:00 <eglynn> o/
15:04:27 <gordc> for anyone who wants to add a bp, start yelling loudly soon or it definitely won't get in.
15:04:57 <gordc> anyone have items related to remaining open bps?
15:06:02 * jasonamyers shakes head
15:06:24 <gordc> i'll take 20s constant silence as nothing... if you have something that needs more than 20s typing, type jibberish first and i'll hold.
15:06:33 <gordc> next!
15:06:36 <gordc> recurring: ceilometerclient release?
15:06:42 <gordc> #topic recurring: ceilometerclient release?
15:07:26 <gordc> i think the only thing we added was aodh support which i don't believe we're ready for yet... cdent did you have aodh gating?
15:07:45 <gordc> with integration tests.
15:07:57 <cdent> there's been work by both me and sileht to get some aodh gating
15:08:09 <gordc> i've seen some check experimentals floating around
15:08:22 <cdent> given the state of the gate it has been hard to track the check experimental stuff that I've been trying (to make sure the infra change is kosher)
15:08:32 <gordc> cool cool
15:08:43 <cdent> I've yet to see the results of one
15:08:48 <sileht> I have write a full autoscaling scenario with gabbi https://review.openstack.org/#/c/207137/
15:09:00 <cdent> that ^ is pretty darn sweet
15:09:01 <gordc> we'll just hold off on ceilometerclient release for now... let me know if you need new release next week.
15:09:03 <sileht> That can be run on aodh/gnocchi/ceilometer repository
15:09:18 <gordc> sileht:  it doesn't need new ceilometerclient rihgt?
15:09:22 <gordc> right*
15:09:35 <sileht> gordc, nop it uses the server side redirection
15:09:45 <gordc> sileht: cool cool.
15:10:03 <gordc> i guess nothing else is needed for ceilometerclient.
15:10:19 <gordc> #topic recurring: Gnocchi status
15:10:41 <gordc> ityaptin: i'm just starting to test the influxdb patch
15:10:49 <gordc> ityaptin: is that all ready for review?
15:11:22 <gordc> cdent: sileht: any other items? or just waiting on the ceilometer dispatcher fixes to get in.
15:11:42 <cdent> there's ongoing test, debug, fix cycles going on
15:11:55 <sileht> nothing except bugs fixing
15:12:01 <_nadya_> gordc: afaik, yes. pinging Ilya now
15:12:13 <gordc> _nadya_: thanks!
15:12:18 <ityaptin> gordc: hi!
15:12:35 <cdent> it's pretty clear there are going to be bugs and missing features that we continue to discover during this period of increased "let's try it in devstack!"
15:12:53 <gordc> ityaptin: ^ few lines up... is the influx driver ready for testing
15:12:55 <ityaptin> gordc: Yes, most part of gnocchi functional is implemented.
15:12:59 <gordc> :)
15:13:06 <sileht> I have discovered a issue with gnocchi:
15:13:15 <gordc> :(
15:13:30 <sileht> https://bugs.launchpad.net/gnocchi/+bug/1479429
15:13:31 <openstack> Launchpad bug 1479429 in Gnocchi "aggregation: issue when one of the resource is no more updated" [Undecided,New]
15:13:31 <uvirtbot> Launchpad bug 1479429 in gnocchi "aggregation: issue when one of the resource is no more updated" [Undecided,New]
15:13:32 <uvirtbot> Launchpad bug 1479429 in gnocchi "aggregation: issue when one of the resource is no more updated" [Undecided,New] https://launchpad.net/bugs/1479429
15:13:40 <sileht> I have not yet a solution
15:14:16 <sileht> it's a bad behavior of 'across metric aggregation' code
15:14:18 <gordc> sileht: is that driver specific or will affect all drivers no matter what?
15:14:30 <gordc> ie only canonical driver?
15:14:44 <sileht> gordc, it's something that need to be fixed in carbonara
15:15:08 <gordc> kk so it theorectically should not affect ityaptin's influxdb driver
15:15:19 <sileht> I dont' think so
15:15:32 <gordc> cool cool. hopes not
15:15:44 <gordc> ityaptin: you might just want to watch out for that bug anyways
15:16:07 <gordc> ityaptin: do you know what is missing from influxdb driver currently?
15:16:21 <gordc> or is it too long to list?
15:17:17 <ityaptin> 1. It doesn't support metricd, we process ts data in realtime
15:17:18 <cdent> jd__ reported that some tests are still failing in his test runs using it
15:17:30 <cdent> but that might be caught up by now
15:17:49 <ityaptin> 2. cross_metric_measures is not implemented yet
15:18:25 <ityaptin> 3. Some tests failing caused by sqlalchemy unexpected (by me) behavior
15:19:38 <gordc> ityaptin: kk. are any of those issues blocked?
15:21:16 <gordc> ityaptin: let us know i guess. or if you need resources to work on the remaining issues.
15:21:48 <gordc> anything else on gnocchi?
15:22:02 <ityaptin> gordc: cross_metric_measures is in progress and jd is looking to sqlalchemy issue
15:22:16 <gordc> ityaptin: awesome!
15:22:39 <gordc> let's move on and revisit later if needed.
15:22:44 <gordc> #topic to set pre-define choices or not to set pre-define choices
15:23:04 <gordc> thorst1: i don't know if you want to explain this. or i can
15:23:16 <thorst1> Sure, I can give a bit of background
15:23:20 <gordc> :)
15:23:49 <thorst1> A recent change ( https://review.openstack.org/#/c/203319/ ) limits the compute inspectors to JUST what is specified in the list.
15:23:56 <thorst1> currently that list is just what is in tree.
15:24:26 <thorst1> Moving forward, Ceilometer is going to a plugin/entry point approach.  But every addition would still require an in-tree change in ceilometer to support that
15:25:04 <thorst1> So the question is - is that the right approach?  It would allow Ceilometer to state what is 'blessed', but for operators/developers looking to make their own they would be limited.
15:25:50 <gordc> so the choices are to continue and Ceilometer is a validator of drivers (even those out of tree), or just let anyone plug in
15:26:34 <thorst1> One question I had while I was thinking about this...  Do we think that there are operators writing custom inspectors for their clouds?
15:26:43 <gordc> thorst1: tbh, since we talked i sort of swayed and i don't see much value in curating list of validated drivers
15:26:43 <_nadya_> hmm, "blessed" means "what is better libvirt or vmware" ?
15:27:16 <gordc> _nadya_: blessed == validated == it meets some criteria
15:27:33 <_nadya_> gordc: ok, I see
15:28:11 <gordc> thorst1: probably not but who knows... i guess that was the point of plugin system
15:29:18 <gordc> llu might have a better idea as he's the compute person... but he's not here today.
15:29:33 <gordc> it seems like no one has a strong opinion.
15:29:45 <gordc> ildikov: ^ if you do.
15:30:07 <thorst1> well, I think its better to move to an approach of listing them in the help...but with plugins, I don't think it makes sense to enforce so strongly
15:30:38 <cdent> I think we should list them in the core code at all
15:30:44 <cdent> shouldn't!
15:30:45 <cdent> sorry
15:30:55 <gordc> thorst1: agreed. we can just say "these are known inspectors"
15:30:59 <cdent> because otherwise it gets out sync
15:31:22 <gordc> cdent: hmm
15:31:23 <cdent> if we do list, then we should list the core ones (in the help string) but also make it clear that external ones are possible
15:31:46 <thorst1> could say something like that and to refer to the wiki for the current list?
15:32:14 <thorst1> for the current *known* list that is
15:33:06 <thorst1> if that approach is OK, I can take the TODO to propose that change?
15:33:37 * cdent looks at gordc
15:33:42 <cdent> is cool to me
15:33:45 <gordc> thorst1: let's maybe start with just listing it in help for now... and if someone complains on patch, we can think about wiki
15:33:57 <thorst1> cool
15:34:29 <gordc> i don't want to maintain wiki... or it should be generally applied to all driver options
15:34:48 <gordc> thorst1: thanks for bringing it up
15:34:55 <thorst1> thank you!
15:35:07 <gordc> cool, next topic.
15:35:24 <gordc> #topic M Midcycle: Dublin need to get dates picked (https://wiki.openstack.org/wiki/Meetings/Ceilometer/M*_Mid-Cycle)
15:35:31 <gordc> i assume this is jasonamyers
15:35:42 <jasonamyers> YES!
15:35:57 <jasonamyers> I really want to have a in person midcycle this time
15:36:07 <jasonamyers> I won't be able to make Tokyo
15:36:29 <jasonamyers> so I'm hoping that we can get together in Dublin in late November early December
15:36:32 * cdent wonders if travel grants can be done on a nomination basis
15:37:06 <jasonamyers> I have a pool up at http://doodle.com/hmukqwzvq7b54cef
15:37:06 <gordc> jasonamyers: expense tokyo as development machine.
15:37:11 <jasonamyers> haha gordc
15:37:23 <jasonamyers> or a poll whatever
15:37:33 <cdent> actually I agree with jasonamyers: I think the midcyle is likely more important than tokyo in terms of getting real work moving
15:37:38 <jasonamyers> anyway If you can come get together sometime during that window it would be awesome
15:37:54 <jasonamyers> Cisco is going to host and sponsor foods etc
15:38:23 <jasonamyers> I'm open to moving in to late November it's just hard to avoid thanksgiving in the US
15:38:34 <gordc> jasonamyers: are we targeting 3 days?
15:38:37 <gordc> or 2?
15:38:43 <jasonamyers> whatever people want to do
15:38:58 <jasonamyers> I'm going to go for a week
15:39:14 <gordc> eglynn: you remember how long paris midcycle was? 2?
15:39:15 <jasonamyers> so if people want to meet for a few days and hack for a few we can do that
15:39:36 <jasonamyers> It's important to have some dates by Aug 18th for ildikov
15:39:41 <jasonamyers> if I remember correctly
15:40:15 <eglynn> 3 days long IIRC
15:40:44 <gordc> eglynn: yeah. seems like it... let's work around assumption it's 2 or 3 days long. so you can budget
15:41:03 <gordc> errr second part of above sentence was general ^
15:41:31 <gordc> cool. thanks for the info and offering to host jasonamyers
15:42:15 <gordc> i'll take it that there's no more questions about midcycle.
15:42:34 <gordc> #topic promote tokyo topics & open discussion
15:42:50 <gordc> we did this last time but feel free to blast your tokyo summit talks now...
15:42:57 <gordc> no guarantee anyone will read or vote.
15:43:07 <thorst1> Power team has a few that we've proposed  :-)
15:43:08 <thorst1> https://www.openstack.org/summit/tokyo-2015/vote-for-speakers/presentation/4300
15:43:31 <thorst1> and a multi hypervisor one as well: https://www.openstack.org/summit/tokyo-2015/vote-for-speakers/presentation/4287
15:43:46 <cdent> it's hard to beat "POWER TEAM" as a team name
15:44:03 <jasonamyers> haha
15:44:05 <gordc> cdent: ibm has good branding.
15:44:14 <thorst1> :-)
15:45:04 <gordc> talk about declarative/pluggable ceilometer: https://www.openstack.org/summit/tokyo-2015/vote-for-speakers/presentation/4685
15:45:25 <gordc> talk about performance testing: https://www.openstack.org/summit/tokyo-2015/vote-for-speakers/presentation/5102
15:45:32 <gordc> and that's my self plug
15:46:07 <prad> me and Ildiko's : https://www.openstack.org/summit/tokyo-2015/vote-for-speakers/presentation/4446
15:46:19 <gordc> there's quite a few ceilometer talks if you search on it... from various users which might be interesting for devs to check out.
15:46:26 <prad> Aodh one from jd https://www.openstack.org/summit/tokyo-2015/vote-for-speakers/presentation/4060
15:47:01 <gordc> cool cool. so the rest is open discussion now
15:47:23 <gordc> not sure if anyone has additional items to raise...
15:47:55 <jasonamyers> I just wanna send a word of thanks you all for helping me get the blueprint code in
15:48:09 <jasonamyers> I know it was annoying to have so much hand holding, but I really appreciate it
15:48:28 <gordc> jasonamyers: thanks for the work
15:49:00 <gordc> just so you know, if it's broken or anything around it is broken, you are now first to blame.
15:49:17 <gordc> basically anything related to pollsters is now on you.lol
15:49:23 <jasonamyers> HA
15:49:25 <jasonamyers> but sure
15:49:30 * cdent dodges a bullet
15:49:51 <gordc> last person who touches it.
15:49:52 <cdent> I was sure that no transform thing made everything my fault
15:50:20 <gordc> your batch patch is not in yet. so i guess it's back on you.
15:50:40 <gordc> anything else?
15:51:28 <gordc> we can continue to git blame at openstack-ceilometer.
15:51:32 <gordc> thanks folks
15:51:35 <gordc> #endmeeting