15:00:24 <witek> #startmeeting monasca
15:00:25 <openstack> Meeting started Wed Dec  5 15:00:24 2018 UTC and is due to finish in 60 minutes.  The chair is witek. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:26 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:28 <openstack> The meeting name has been set to 'monasca'
15:00:36 <witek> Hello everyone!
15:01:25 <dougsz> Hey all
15:01:32 <joadavis> hello
15:01:39 <witek> hi dougsz and joadavis
15:02:18 <witek> I've created last minute agenda
15:02:26 <witek> https://etherpad.openstack.org/p/monasca-team-meeting-agenda
15:02:50 <witek> #topic Events Listener
15:02:56 <witek> https://review.openstack.org/583803
15:03:03 <joadavis> Thanks again for the reviews
15:03:20 <joadavis> I haven't groked all the nova cells concepts yet
15:03:39 <witek> we have some constructive discussion in review
15:04:05 <dougsz> I am pretty sketchy myself - I've mostly read the docs so far
15:04:58 <dougsz> It's something we are adding support for in Kolla, and it would be nice to make sure it works well with the Events listener
15:05:12 <witek> yes, agree
15:06:33 <joadavis> For the first implementation, if we can get all the messages sent to the kafka monevents topic that may work
15:07:17 <joadavis> I don't know if we need to separate events based on which cell they come from. Assuming there is some data in the dimensions that would identify the source cell for the event
15:08:14 <witek> the issue dougsz is bringing up is, that there might be no connectivity between Cell message queue and Monasca cluster
15:08:46 <joadavis> I had thought we could just use existing code for connecting to rabbitmq, but will need to do a quick check if it can handle multiple rabbitmq connections.  But if there is no network connection allowed, that is a problem.
15:09:56 <joadavis> ideas for options if there is no connectivity?
15:11:26 <dougsz> I was thinking this might be a use case where posting events to the Monasca Events API would work well
15:13:00 <joadavis> yeah, that might be the way. Would we need a monasca service in each cell to listen to the cell's RabbitMQ for events then post them to the Events API?
15:13:13 <witek> I'd suggest to seek some more expert knowledge on Nova Cells
15:13:24 <dougsz> That was the way I imagined it could work.
15:13:25 <joadavis> monasca service or modified monasca agent
15:13:26 <witek> how they are supposed to be deployed
15:13:38 <joadavis> yes
15:14:02 <dougsz> We might be able to get some info from Cern
15:14:07 <dougsz> I will ask around
15:14:12 <joadavis> thanks
15:14:15 <witek> good idea, thanks
15:15:02 <witek> let's continue the discussion in review
15:15:36 <witek> the last unclarified question seems to be, how the listener is supposed to be delpoyed
15:16:18 <witek> as part of API, or as separate new Monasca service
15:16:52 <witek> no, actually not the only one
15:17:08 <witek> should it send to Kafka, or to Events API?
15:18:22 <witek> #topic Python 3 support
15:18:34 <joadavis> sending to Kafka seems more efficient.  Though we can use the API if needed for cells
15:19:06 <joadavis> I took a quick look at the py3 cassandra failure you mentioned in the agenda
15:19:27 <joadavis> That is a strange tempest error, and not obviously related to cassandra
15:19:52 <joadavis> but I pinged sgrasley and jgu to see if they know of any issues with cassandra clients and py3
15:20:43 <witek> adriancz has mentioned issues with string handling
15:21:18 <witek> he has set up CI jobs for running tempest tests with Python 3
15:21:22 <witek> https://review.openstack.org/619987
15:22:47 <witek> the jobs are marked as non-voting for now, but we should fix them as soon as possible
15:23:02 <witek> thanks joadavis for forwarding this
15:24:04 <joadavis> agreed. (though I wish all zuul tests were more predictable - we still get many random failures)
15:24:19 <joadavis> (but I'll stop complaining now)
15:24:50 <witek> :)
15:25:24 <witek> I have observed long queues of Zuul jobs recently
15:26:00 <witek> at some point it has taken several hours before the jobs were even started
15:26:14 <joadavis> I tried initiating 3 rechecks over in ceilometer repos yesterday for different random failures
15:26:57 <dougsz> :(
15:27:12 <witek> did it improve anything?
15:27:40 <joadavis> at least one passed (it was the mailing list change).  I need to read my email and see about the rest
15:28:33 <joadavis> I think something changed in Zuul. in the past when you did a recheck it would post back a message indicating it started the gates, but now I don't see one until it finishes
15:29:23 <witek> but you can check the status on http://zuul.openstack.org/status
15:29:31 <joadavis> yes
15:29:49 <witek> let's move on
15:30:06 <witek> #topic project health (feedback for TC)
15:30:32 <witek> gmann has contacted me and asked for feedback regarding Monasca project health
15:30:52 <witek> TC tries to track potential issues in projects
15:31:01 <witek> https://wiki.openstack.org/wiki/OpenStack_health_tracker
15:31:48 <witek> he has asked among others if there is anything TC could do to support our project?
15:33:18 <witek> please let me know, if you have ideas
15:33:48 <witek> #topic Trove + Monasca + Vitrage + Mistral
15:33:50 <joadavis> For us, I think we have a relatively healthy community, though would always like more contributors (as all the projects say)
15:34:25 <witek> joadavis: agree, we have very few active contributors
15:34:26 <joadavis> I think the TC could help in coordinating and supporting other services that consume monasca data, such as watcher.  Or vitrage
15:34:42 <joadavis> and the SIGs help with that too
15:34:52 <joadavis> and that transitions to our next topic. :)
15:36:29 <witek> thanks joadavis, I thinks we've said something similar to dhellmann during the Denver PTG
15:37:08 <joadavis> dumb question: is Trove still around?  I hadn't heard about it in a year or so
15:37:32 <witek> OK, let's move to the Trove topic
15:37:48 <witek> I think the project was sort of abandoned
15:37:57 <witek> and then Samsung jumped in
15:39:01 <witek> they have given a presentation in Berlin
15:39:11 <dougsz> I recall Bartosz Żurkowski asking about using Monasca to monitor it before the Berlin summit
15:39:14 <witek> https://www.openstack.org/summit/berlin-2018/summit-schedule/events/22273/towards-production-grade-database-as-a-service-in-openstack
15:39:20 <dougsz> ( a Trove core)
15:39:31 <dougsz> Not sure where he got to
15:39:48 <dougsz> Ah, you just answered that :)
15:40:00 <witek> I have missed that presentation
15:40:32 <witek> but they presented POC for integration between Trove, Monasca, Vitrage and Mistral
15:40:42 <joadavis> cool
15:41:10 <witek> a draft for Monasca datasource in Vitrage has been pushed just today
15:41:18 <witek> https://review.openstack.org/622899
15:41:52 <witek> Ifat has reported this in the self-healing meeting today
15:42:20 <witek> and they are willing to help to get this merged
15:43:16 <joadavis> Do they need any work on the Monasca side for this?  other than configuring a webhook to publish to Vitrage?
15:43:51 <witek> I think not
15:44:58 <witek> we have discussed an option for sending Monasca notifications via message queue
15:45:09 <witek> which is kind of default in Vitrage
15:45:33 <witek> but I think it's not necessary at the end
15:46:24 <witek> that's all from my side
15:46:46 <witek> I didn't come to merging notification changes
15:47:12 <witek> but I pushed the Alembic devstack change, to have it all complete
15:47:22 <witek> turned out to be pretty easy
15:47:31 <witek> https://review.openstack.org/622361
15:48:03 <dougsz> thanks witek
15:48:19 <joadavis> I filed two new stories yesterday.  One is a simple cleanup of some ceilometer monitoring in the agent.
15:48:43 <joadavis> The other is https://storyboard.openstack.org/#!/story/2004539 and might take a bit more work
15:49:15 <joadavis> I don't know if removing an alarm configuration for http_alive was supported well.
15:49:31 <joadavis> I may follow up with sc on this to see if he has any ideas.
15:50:08 <witek> alarm or collector configuration?
15:50:33 <joadavis> sorry, for the agent collector configuration
15:51:23 <witek> will you or someone from SUSE work on this?
15:51:36 <joadavis> the use case is that a compute host has been removed, so we want to do cleanup.  but at this point, we just know the hostname and not all the target_hostnames used in setting up the configuration
15:51:45 <joadavis> I'm working on it currently
15:51:53 <witek> OK, thanks
15:52:33 <joadavis> I'm sharing in case anyone has insight in to the agent.
15:52:49 <joadavis> otherwise, that is enough from me
15:52:51 <dougsz> sorry, haven't hit this use case yet
15:53:39 <joadavis> we put a lot of work in to getting a cloud up, not as much in to removing parts of it. :)
15:53:47 <witek> :)
15:55:21 <witek> OK, I'll be wrapping up if there is nothing else
15:55:36 <witek> thank you for joining
15:55:50 <witek> and see you next week
15:56:05 <witek> for interested
15:56:05 <joadavis> goodbye
15:56:19 <witek> second round of self-healing meeting starts in one hour
15:56:37 <witek> in #openstack-self-healing
15:56:47 <witek> bye bye
15:56:48 <dougsz> bye all
15:56:53 <witek> #endmeeting