15:00:54 <witek> #startmeeting monasca
15:00:55 <openstack> Meeting started Wed Sep 25 15:00:54 2019 UTC and is due to finish in 60 minutes.  The chair is witek. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:56 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:59 <openstack> The meeting name has been set to 'monasca'
15:01:21 <Dobroslaw> hi
15:01:36 <witek> hello brtknr, sc, Dobroslaw
15:01:52 <brtknr> tthello
15:02:03 <witek> the agenda is very light today:
15:02:06 <witek> https://etherpad.openstack.org/p/monasca-team-meeting-agenda
15:02:20 <witek> so please add topics if you have some
15:02:34 <witek> #topic Release candidate for Train
15:02:35 <brtknr> db per tenant
15:03:02 <witek> brtknr: added to the etherpad
15:03:04 <brtknr> witek: thanks for adding :)
15:03:59 <witek> the release team has created patches for releasing all the remaining repositories
15:04:13 <sc> what do we have to do?
15:05:01 <witek> release liaison or PTL has to confirm these
15:05:12 <witek> most of the changes have been merged already
15:05:24 <sc> anything we are missing to merge?
15:05:49 <sc> tomorrow I'll be available most of the day
15:05:51 <witek> also, stable/train branches have been cut with these releases
15:05:57 <witek> I
15:06:50 <witek> I've kept api and notification on hold as it would be good to include last changes from these repos in Train
15:07:30 <witek> I'm referring to:
15:07:37 <witek> https://review.opendev.org/651249
15:07:44 <witek> https://review.opendev.org/674812
15:08:46 <sc> there was a good reason why I didn't do +2 W+1
15:09:54 <Dobroslaw> Adrian is quite happy with merging apis change
15:11:20 <witek> Dobroslaw: have you been testing this change, I've seen your +1 on older PS
15:11:23 <witek> ?
15:11:49 <Dobroslaw> unfortunately lately no...
15:12:13 <witek> let's try to push these two in the next days
15:12:36 <witek> I'll have to cut the branch next week latest
15:13:34 <witek> Train release schedule is available here:
15:13:36 <witek> https://releases.openstack.org/train/schedule.html
15:13:59 <witek> with final release on 16 October 2019
15:14:11 <brtknr> is there any possibility of incorporating db per tenant changes?
15:15:00 <witek> checking
15:16:21 <witek> the change in release repo hasn't merged yet
15:16:43 <witek> let's talk about it in a minute
15:17:04 <brtknr> okay
15:17:32 <witek> are there any questions about release schedule?
15:18:16 <witek> #topic Ussuri planning meeting
15:18:31 <witek> thanks a lot for filling in the survey
15:18:38 <witek> https://doodle.com/poll/ugxr89tqmkfwa5r7
15:19:42 <witek> currently, Oct 16-17 seem to be the best days
15:20:57 <witek> if nothing changes, I'll post to the list with these two days
15:21:24 <witek> and we can start collecting topics and features for the next cycle
15:21:36 <witek> I'll update the etherpad for that
15:22:10 <witek> do these dates work for everyone?
15:22:28 <sc> they work for me
15:22:38 <witek> nice
15:23:21 <witek> #topic InfluxDB database per tenant
15:23:39 <witek> https://review.opendev.org/#/q/status:open++branch:master+topic:story/2006331
15:24:13 <witek> brtknr: I've reviewed the persister change today
15:24:57 <witek> have you evaluated the impact on write throughput?
15:25:30 <brtknr> witek: yes I saw, dougsz is not here atm, I haven't suffered much from write performance, mostly read... this change is to mostly mitigate that
15:25:55 <brtknr> but i can see this being a problem where there are large number of projects all posting metrics to the same database
15:27:01 <brtknr> witek: do you have a suggestion for a method to measure write throughput?
15:28:36 <witek> there are some scripts available in https://github.com/monasca/monasca-perf
15:29:01 <witek> but I'm not sure how relevant they are, many might be obsolete
15:32:42 <brtknr> witek: so you're mostly okay with the migration tool and the read side of things?
15:33:32 <witek> I've started with the persister, the code is clean and nice, I'm only worried about increasing number of requests to InfluxDB
15:34:00 <witek> I'll think about it a little longer
15:34:16 <witek> let me get back to it tomorrow
15:34:57 <brtknr> is it feasible to have lots of kafka topics?
15:35:04 <brtknr> e.g. one per tenant?
15:36:13 <witek> I think it might be a good approach, but we'd have to introduce management of Kafka topics
15:36:34 <brtknr> in a single tenant case, the perfomance is the same as before... when there are two tenants, the perforance will depend on how interleaved the activity is
15:36:50 <brtknr> more interleaved, less performant
15:37:40 <witek> would your change affect performance with option `db_per_tenant` disabled?
15:38:38 <brtknr> witek: with db_per_tenant disabled, it will be the same outcome as before
15:39:26 <brtknr> witek: actually, i am wrong there
15:40:11 <brtknr> It would be a simple change to revert to former behaviour when db_per_tenant is disabled
15:40:37 <witek> yes, see you point
15:40:58 <witek> let me get back to it tomorrow
15:41:23 <witek> another question, how important is it for you to include in stable/train branch?
15:43:34 <brtknr> we are keen to get it in but not if you dont think its quite ready
15:43:50 <brtknr> we were aiming for stable/train
15:45:23 <witek> are you building from official stable releases, or have some downstream process for building deliveries?
15:45:59 <witek> if you strongly rely on stable/train release I'd try to push it in
15:48:06 <witek> anyway, I've blocked cutting the release for now
15:48:24 <witek> let's get back to it tomorrow
15:50:01 <witek> do we have any other topics for today?
15:50:42 <joadavis> Not much from me, just my list of long running topics.
15:51:22 <witek> go on, we have 10 minutes :)
15:51:59 <joadavis> I recently looked at monasca events listener again.  The ceilometer events code may not be very suitable for our use, so I started looking at vitrage for an example
15:52:36 <joadavis> basing something on the current version of oslo_messaging is likely the best option
15:52:55 <witek> Dobroslaw, adriancz: do you remember if we had some prototype code for events agent?
15:53:19 <Dobroslaw> I remember only api part
15:54:05 <Dobroslaw> don't know about agent
15:54:15 <witek> api is merged, but I think we also had some code created during investigation, not sure of that though
15:54:54 <brtknr> witek: sorry, I meant to answer earlier... we mostly use upstream releases, we'd usually rather not maintain our own downstream process.
15:55:35 <joadavis> brtknr: that is a good approach. :)
15:55:49 <witek> joadavis: do we need to update the spec
15:55:52 <witek> brtknr: thanks
15:56:26 <joadavis> so far the spec is ok.  If we get a prototype we can adjust the spec to match
15:56:55 <joadavis> I'd like to work more on it, but have a lot of other things with higher priority
15:57:31 <joadavis> (and likely should bump the monasca-ceilometer publisher submission to ceilometer first)
15:59:43 <witek> let's prepare some material about events agent for the planning meeting, so that we can prioritize it for the next cycle
16:00:19 <witek> I have to wrap up, thanks for joining and the discussion
16:00:32 <joadavis> thanks all
16:00:33 <witek> see you next week, bye
16:00:38 <witek> #endmeeting