15:01:26 <gordc> #startmeeting telemetry
15:01:27 <openstack> Meeting started Thu Mar  3 15:01:26 2016 UTC and is due to finish in 60 minutes.  The chair is gordc. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:01:28 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:01:32 <openstack> The meeting name has been set to 'telemetry'
15:01:54 <liusheng> o/
15:02:06 <_nadya_> o/
15:02:11 <r-mibu> o/
15:02:17 <llu-laptop> o/
15:02:33 <ildikov> o/
15:02:51 <gordc> ok let's get this started.
15:02:55 <idegtiarov> 0/
15:03:04 <gordc> #topic recurring: roadmap items (new/old/blockers) https://wiki.openstack.org/wiki/Telemetry/RoadMap
15:03:26 <gordc> i just cut m-3 this morning against HEAD
15:03:42 <gordc> first. thanks for all the work you folks have done this cycle
15:04:13 <gordc> regarding FFE, right now i'm just tracking idegtiarov timedelta patch
15:04:24 <gordc> and r-mibu's tempest work (if it finishes)
15:04:51 <ildikov> gordc: the tempest work in Aodh?
15:04:59 <gordc> if someone has anything outstanding, please yell now (although it probably won't work)
15:05:10 <gordc> ildikov: ceilometer and aodh
15:05:21 <gordc> r-mibu: do you still think you'll have time?
15:05:37 <r-mibu> yep, i'll finish in this cycle
15:05:59 <ildikov> gordc: ok, cool
15:06:06 <gordc> r-mibu: ok. just so you know, it ends very soon :)
15:06:12 <r-mibu> i'm planning to use different name in our plugins to avoid param conflict
15:06:24 <liusheng> I will try to help
15:06:25 <r-mibu> so, it can be finish only in our repos,
15:06:29 <ildikov> gordc: I started to check the Aodh one, I just got confused by the WIP sign
15:06:43 <r-mibu> liusheng, thanks!
15:06:55 <ildikov> r-mibu: cool, thanks for the clarifications
15:07:27 <gordc> ok. so final cut is in ~1 week
15:07:31 <r-mibu> i will upload patches tomorrow, i'm working in my lab machine now...
15:07:33 <gordc> Mar 14-18
15:07:52 <_nadya_> gordc: bug fixes can still be merged, right?
15:07:57 <gordc> r-mibu: ok. thanks. hopefully we can merge it in.
15:08:29 <r-mibu> ok, apologies for the delay :-p
15:08:37 <gordc> _nadya_: yes, for next few weeks please test like crazy.
15:08:46 <gordc> r-mibu: not a problem.
15:08:50 <ildikov> gordc: final cut means FFE deadline, right?
15:08:52 <_nadya_> gordc: yep...
15:09:34 <gordc> ildikov: yeah, rc1. i'm going to make executive decision and say if you can't get feature in for RC1, we'll push it to N*
15:09:47 <gordc> ildikov: seems like the logical answer.
15:09:59 <ildikov> gordc: yeah, right, this is what I meant
15:10:03 <ildikov> gordc: tnx!
15:10:39 <gordc> so yeah, everyone with +2/+A powers, please do not merge anything unless it's: https://review.openstack.org/#/c/286469
15:10:54 <gordc> or https://review.openstack.org/#/q/owner:r-mibu%2540cq.jp.nec.com+status:open
15:11:25 <gordc> or a bug
15:11:27 <ildikov> gordc: -2ing other feature patches until N?
15:12:07 <gordc> ildikov: i've -2 a few. feel free to -2 if you like, but just remember you need to unblock it in a few weeks.
15:12:19 <ildikov> gordc: I mean like what other projects are doing that "block this until next release" or we don't have that many?
15:12:26 <gordc> ildikov: i would say just leave it unless you feel like -2'ing stuff. :)
15:12:39 <gordc> ildikov:  i trust most of us to not merge stuff.
15:12:45 <ityaptin-tablet> O/
15:12:54 <ildikov> gordc: ok, works for me, I'm not that eager with -2ing
15:13:03 <gordc> but i guess if you think it's a feature pretending to be a bug, -2 it
15:13:33 <ildikov> gordc: it just can make it easier to remember that we are still in frozen state :)
15:14:07 <ildikov> gordc: ok, cool, crystal clear, tnx :)
15:14:31 <gordc> ildikov: cool cool.
15:14:39 <_nadya_> ok
15:15:11 <gordc> let's do a quick run through of projects (i don't think we have much to discuss)
15:15:15 <gordc> #topic aodh topics
15:15:27 <gordc> just a heads up, i'm going to defer aodhclient in heat
15:15:38 <gordc> i didn't get a response when i asked about FFE last week
15:15:49 <gordc> and i also realised it's a crazy big change.
15:16:17 <gordc> if someone wants to work on heat+aodhclient, here's the patch: https://review.openstack.org/#/c/284985/
15:16:23 <ildikov> then it sounds reasonable to defer
15:16:33 <gordc> feel free to take it over (i only got unit tests working)
15:17:07 <liusheng> gordc: really a big change!
15:17:33 <gordc> liusheng: yeah, it was more work than i wanted
15:18:17 <gordc> liusheng: do you want to talk about 'aodh search --query' vs 'aodh list --query'?
15:18:22 <gordc> we can defer to ML as well
15:18:40 <_nadya_> gordc: this change goes to N?
15:18:53 <liusheng> gordc: yeah, we can have a shot discussion
15:19:00 <gordc> _nadya_: aodhclient support in heat?
15:19:03 <liusheng> https://review.openstack.org/#/c/283958/
15:19:06 <_nadya_> gordc: yes
15:19:30 <gordc> yeah, i'm not working on it. and i don't think heat wants to do it (no one replied when i asked about it last week)
15:20:00 <_nadya_> gordc: I think I will be able to work on this soon
15:20:30 <gordc> _nadya_: sure if you like. feel free to takeover my patch (or start from scratch)
15:20:42 <_nadya_> gordc: ok, cool
15:21:14 <gordc> #topic aodh - list vs search
15:21:29 <gordc> liusheng: go for it
15:22:13 <liusheng> gordc: ok, I have a change try to drop the mandatory of "--type" in alarm list command
15:22:27 <liusheng> do you have any thoughts about that ?
15:22:32 <liusheng> https://review.openstack.org/#/c/283958/
15:23:14 <gordc> i don't mind dropping mandatory type. i think my question is what we want to do with list vs search
15:23:32 <gordc> want to see if anyone has an opinion on this.
15:24:08 <_nadya_> liusheng: what's the reason of dropping --type?
15:24:14 <ildikov> does list with query equal to search
15:24:19 <ildikov> ?
15:24:26 <gordc> disclaimer: the format was taken from gnocchiclient
15:24:29 <liusheng> yes, support query filters in "aodh alarm list" or support complex query in alarm search
15:24:50 <liusheng> in ceilometercliet, we supportted both
15:25:46 <liusheng> _nadya_: I think it is not reasonable to force users specifying --type if they just want to list alarms
15:26:37 <_nadya_> liusheng: I see, agreed
15:26:42 <liusheng> _nadya_: and, may be users want to get all the alarms in "ALARM" state and regardless the type
15:27:30 <_nadya_> liusheng: so the question "what filters should be supported in alarm list and alarm search"?
15:28:58 <liusheng> _nadya_: no, we have support almost all the query filters by "alarm search" (complex query), so gordon think we don't need duplicated functionaities in "alarm list"
15:29:23 <gordc> _nadya_: so in gnocchiclient, list doesn't really have any filters (just how much details are shown). search is what you use to filter
15:29:55 <gordc> liusheng: i think it's confusing to have query in both list and search. we should unify them.
15:30:07 <gordc> either a) move all filtering capability to search
15:30:11 <liusheng> but it has been already supportted in aodh-api :)
15:30:18 <gordc> or b) drop search and move everythin to list
15:31:06 <gordc> right so there will be --query and --complex-query options... but they should both be under same command.
15:31:23 <gordc> it just seems confusing to have list --query and search --query
15:31:28 <gordc> or maybe that's just me.
15:31:43 <ildikov> I tend to agree that it causes confusion
15:31:57 <ildikov> it looks momre reasonable under search
15:32:09 <ildikov> s/momre/more/
15:32:09 <liusheng> gordc: I thinks it is better to support filters in alarm list, maybe we can support them both temporally ?
15:32:37 <gordc> liusheng: drop alarm search command then?
15:32:37 <ildikov> what if we call query as filter for alarm list?
15:32:48 <_nadya_> got it. IIUC, "list" and "search" both return the list of results. Tend to agree that these commands look very similar
15:33:02 <gordc> _nadya_: right
15:33:09 <liusheng> gordc: since we have plan to support pagination in the future, yes, I think pagination and filtering query is enough
15:33:13 <ildikov> and keep complex-query for search or pick one and stick with that
15:33:15 <liusheng> what do you think ?
15:34:15 <gordc> are we dropping alarm search command? i don't mind this.
15:34:40 <gordc> i think right now it's not obvious what diff between list --query|filter vs search --query is...
15:35:40 <ildikov> I would drop list, but I think it's just a matter of taste :)
15:36:00 <liusheng> gordc: I prefer that, but I respect all of your opinions
15:36:13 <gordc> :)
15:36:39 <gordc> we should drop one, or leave all filtering (complex and regular) capabilities in search
15:37:13 <gordc> gnocchiclient uses second approach
15:37:31 <gordc> i think most clients for other projects, use first approach.
15:37:41 <ildikov> do we have any intention to keep these clients similar?
15:37:51 <liusheng> agree, if we just support "alarm list" without any parameter, it is equal to "alarm search" with no parameter specified
15:37:59 <ildikov> I mean what is under Telemetry umbrella
15:38:14 <_nadya_> +1 to drop one. but the question is in backward-compatibility. Do we care about it? or aodhclient is a new one and we don't want to mantain ceilometer alarming there?
15:38:21 <gordc> ildikov: no plans. i just happened to build aodhclient off gnocchiclient becuase it was using all the latest libs
15:39:01 <gordc> _nadya_: i think we can ignore backward-compatbility (for now) :)
15:39:29 <ildikov> gordc: sure, I meant that you referred to it as it uses the second approach so we might keep aodhclient similar so that who uses both knows what to expect
15:39:58 <liusheng> for  backward-compatibility, we need support both :)
15:40:29 <gordc> ildikov: right. or if you use one, you get get confused.lol
15:40:53 <ildikov> hmm, maybe we can still do things now quickly and tell people that they don't remember correctly that aodhclient supported this or that... ;)
15:41:13 <ildikov> gordc: why would I use only one? ;)
15:42:44 <gordc> liusheng: i don't have a recommendation, maybe defer to ML?
15:42:57 <gordc> unless someone has a strong opinion righ tnow
15:43:04 <gordc> right now*
15:43:21 <liusheng> gordc: OK, I will open a thread in ML
15:43:32 <gordc> liusheng: thanks!
15:43:39 <gordc> #topic ceilometer topics
15:43:42 <_nadya_> yep, it would be better to discuss it in ML
15:43:43 <gordc> anything here?
15:43:56 <ifat_afek> hi
15:44:25 <ifat_afek> you switched to ceilometer topic, but can I take 2 min to talk about aodh-vitrage integration? or should I do it in ML?
15:44:32 <gordc> sure.
15:44:43 <gordc> we have nothing for ceilometer anyways
15:44:54 <gordc> #topic aodh+vitrage integration
15:45:03 <gordc> ifat_afek: go for it
15:45:03 <ifat_afek> so we started working on aodh-vitrage integration. basically, it has 3 parts:
15:45:11 <ifat_afek> 1. vitrage should periodically query aodh alarms. this seems quite straight forward
15:45:19 <ifat_afek> 2. vitrage should be notified whenever an aodh alarm state is changed. this is more complicated, and we decided to implement it at a later stage
15:45:24 <ifat_afek> 3. vitrage should trigger deduced alarms (vitrage alarms) in aodh. this is also not trivial
15:45:35 <ifat_afek> I created two blueprints, for #1 and #3. I will be happy to get your feedback about them.
15:45:42 <ifat_afek> https://review.openstack.org/#/c/287686
15:45:49 <ifat_afek> https://review.openstack.org/#/c/287861
15:46:03 <ifat_afek> there is much more information in vitrage wiki page: https://wiki.openstack.org/wiki/Vitrage
15:47:15 <gordc> ifat_afek: cool cool, we can comment on the specs
15:47:34 <gordc> i think #2 is possible since we have zaqar notifier but i haven't really tried
15:47:35 <ifat_afek> gordc: that would be great, thanks
15:48:02 <ifat_afek> we do plan to implement #2, but for now we will focus on 1 and 3
15:49:07 <gordc> cool cool, so i guess aodh devs can comment there.
15:49:08 <liusheng> ifat_afek: hi, thanks, what is main reason for periodically querying aodh alarms ?
15:49:21 <_nadya_> ifat_afek: will vitrage create it's own alarms in aodh?
15:49:55 <ifat_afek> liusheng: we would like to have all aodh alarms in vitrage for rca and deduced alarms calculations. and doing periodic query seems easier than getting notifications about changes...
15:50:17 <ifat_afek> _nadya_: we would like to do this. this is the purpose of the #3
15:51:50 <liusheng> ifat_afek: will look the specs, thanks :)
15:52:02 <ifat_afek> liusheng: great, thanks!
15:52:25 <gordc> cool cool. open discussion?
15:52:30 <gordc> #topic open discussion
15:52:57 <gordc> just an fyi, ~1 week until PTL self-nominations if anyone is interested.
15:53:06 <gordc> that's all i have
15:53:18 <liusheng> do you have any plan about the Bug smash ? :-)
15:53:47 <liusheng> https://etherpad.openstack.org/p/OpenStack-Bug-Smash-Mitaka-Bugs
15:53:57 <gordc> liusheng: not formally. we have a lot of bugs though in aodh/ceilometer/gnocchi :)
15:55:04 <liusheng> gordc: :)
15:55:43 <gordc> sorry folks, i have to run to a meeting. we all good?
15:56:20 <gordc> cool cool.
15:56:25 <gordc> thanks everyone again
15:56:27 <gordc> #endmeeting