15:00:43 <witek> #startmeeting monasca
15:00:47 <openstack> Meeting started Wed Sep 19 15:00:43 2018 UTC and is due to finish in 60 minutes.  The chair is witek. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:48 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:50 <openstack> The meeting name has been set to 'monasca'
15:01:01 <witek> hello everyone!
15:01:03 <srwilkers> o/
15:01:05 <dougsz_> hey all
15:01:08 <koji_n> hi
15:01:09 <chaconpiza> o/
15:01:17 <witek> hi srwilkers, nice to see you
15:01:25 <srwilkers> witek: you as well!
15:01:33 <witek> the agenda as usual here:
15:01:37 <witek> https://etherpad.openstack.org/p/monasca-team-meeting-agenda
15:01:52 <joadavis> hi
15:01:54 <witek> quite full today
15:02:04 <witek> let's move on
15:02:16 <witek> #topic fluentd plugin
15:02:48 <dougsz_> So not thing too exciting here. Development stopped on the plugin.
15:02:58 <dougsz_> We forked it, added a couple of fixes.
15:03:16 <dougsz_> And I recently moved it from stackhpc github to monasca github.
15:03:25 <witek> great, thanks for that
15:04:08 <witek> srwilkers: will you be interested in collecting logs with Monasca as well?
15:04:35 <witek> fluentd plugin could be interesting in that case
15:04:38 <srwilkers> witek: for logs for the openstack services, we're using the oslo-log fluentd handler/formatter at the moment
15:05:03 <witek> ok
15:05:05 <srwilkers> then the standard processing/output to elasticsearch
15:05:21 <srwilkers> i havent looked much into what monasca provides from a logging standpoint admittedly
15:06:14 <witek> basically, adding multitenancy to the standard ELK
15:06:30 <witek> and basic alerting
15:06:57 <srwilkers> oh, interesting.  i'll need to do some reading/poking before next week then
15:07:21 <witek> dougsz_: thanks for moving the repo
15:07:31 <dougsz_> If you want a sample fluentd output config file for the above you can ping me on IRC, it's very easy.
15:07:36 <dougsz_> np
15:07:45 <dougsz_> That's it from me on this topic - i see we have lots today!
15:07:59 <witek> #topic ptg summary
15:08:20 <witek> I have put together most important points we talked about last week
15:08:26 <witek> https://wiki.openstack.org/wiki/MonascaSteinPTG
15:08:38 <witek> it's linked from the main wiki page
15:09:01 <witek> please add or update if something missing or incorrect
15:09:37 <witek> I also wanted to give a short update on two topics we talked about in different sessions
15:09:47 <witek> OpenStack Helm being the first one
15:09:48 <joadavis> nice
15:10:30 <witek> srwilkers: could you shortly say, what OSH does and why?
15:11:44 <witek> busy or writing :)
15:12:10 <srwilkers> the tl;dr is openstack-helm seeks to provide a set of curated helm charts to deploy openstack services, along with supporting infra services, on top of kubernetes.  the why is to take advantage of what kubernetes offers, in addition to making the charts flexible in the configuration sense
15:12:42 <srwilkers> ie: exposing the right values to override to customize your deployment to your needs
15:13:10 <witek> thanks, it's mainly driven by AT&T, right?
15:13:49 <dougsz_> Will it use Kolla images or will you add support for a different format?
15:13:55 <srwilkers> for now -- we're always looking for new contributors though :)
15:14:14 <srwilkers> dougsz_: we use loci as the default for the openstack services, but we support kolla images as well
15:14:40 <dougsz_> Good to know
15:14:41 <witek> we don't have LOCI images as of now
15:15:06 <witek> dougsz_ has worked on adding images in Kolla
15:15:13 <srwilkers> awesome :)
15:15:32 <dougsz_> I think the kolla images are basically complete and merged upstream
15:15:50 <dougsz_> We don't have plans for LOCI (only a small shop)
15:15:52 <witek> we also have an ongoing effort of adding lightweight images in monasca repos
15:16:30 <witek> so, in the OSH we talked about adding Monasca to OSH
15:16:55 <witek> srwilkers: wants to lead the PoC
15:17:21 <witek> please let us know, if you need any help
15:17:24 <dougsz_> awesome!
15:17:33 <srwilkers> witek: i certainly will.  i started on the plane home from denver :)
15:17:39 <witek> wow
15:17:39 <srwilkers> hoping to have something to show soon
15:18:06 <witek> how long was the flight? :)
15:18:22 <srwilkers> only two hours -- saint louis isn't far from denver, thankfully
15:18:45 <dougsz_> So the spec talks about Fluent+Elastic+Kibana and Prometheus
15:19:08 <dougsz_> What's inspired deploying Monasca?
15:20:09 <joadavis> I think having a few of us Monasca people show up in the OSH sessions at PTG. :)
15:20:13 <srwilkers> from the beginning, we wanted to provide the ability to use whatever supporting infrastructure an operator may want to use.  monasca would be the first alternative we'd add to show we've accomplished that
15:20:35 <srwilkers> i've also been working to add the elasticbeats as well, which works successfully
15:20:49 <srwilkers> but would like to get the monasca work done first :)
15:21:17 <joadavis> Its pretty cool that OSH can be flexible
15:21:44 <dougsz_> yeah, definitely. Both approaches have their advantages.
15:22:25 <witek> thanks, also timothyb89 could a good contact person, he worked on charts in monasca-helm repo
15:22:38 <witek> I have given him heads-up already
15:22:45 <srwilkers> good to know -- i'll keep his nick handy in case i have any questions
15:23:40 <witek> I have also talked with ricolin (Heat PTL) about autoscaling with Heat and Monasca
15:24:01 <witek> the integration has been implemented some time ago already
15:24:14 <witek> but documentation is missing completely
15:24:35 <witek> we though it would be great if we could review the existing implementation
15:24:46 <witek> and publish the use case in self-healing SIG
15:25:02 <witek> similarly as it happens here:
15:25:03 <witek> https://review.openstack.org/602505
15:25:38 <witek> I think it can improve the visibility of the project
15:26:28 <witek> also, some cross-project testing could be useful
15:26:53 <witek> I have added these two topics to the priority game spreadsheet
15:27:25 <witek> as well as the `monasca-thresh` leftover from the last cycle
15:27:53 <witek> dougsz_, chaconpiza: you may want to reconsider your votes to that respect
15:28:35 <witek> are there any other topics which we should add to the list?
15:29:18 <dougsz_> thanks witek - I noticed those.
15:30:09 <chaconpiza> ok, I'll spread a little bit the vote
15:30:09 <witek> folks, please put your money on the topics until end of week
15:30:21 <dougsz_> Support latest InfluxDB?
15:31:01 <witek> Matthias from our team work on 1.6 already
15:31:14 <dougsz_> awesome!
15:32:17 <witek> ok, let's move on
15:32:29 <witek> #topic value_meta in alarms
15:33:01 <dougsz_> So Charana, our intern has produced some patches for this
15:33:42 <dougsz_> The idea being that you could set an alarm for log error messages, and get notifications with snippets of the error message, for example.
15:33:58 <dougsz_> There were a couple of issues.
15:34:18 <Charana> hi
15:34:31 <witek> hi Charana, nice meeting you
15:34:46 <dougsz_> Firstly, a change to monasca-common has been made, and monasca-thresh gates on tempest which needs the change in common to pass.
15:35:37 <dougsz_> So our focus is to get the patch to monasca-common merged first.
15:35:48 <Charana> the common change supports a value_meta string field that thresh will set when submitting alarms to kafka
15:36:09 <witek> I recall we have talked about it in Dublin
15:36:20 <dougsz_> That's correct
15:36:26 <Charana> its a minor change, https://review.openstack.org/#/c/603099/
15:36:49 <witek> did you think about adding the value to the message as well?
15:37:25 <Charana> theres a patch for monasca-notification to include this in the slack notification https://review.openstack.org/#/c/603127/
15:37:31 <dougsz_> We could. For our use case, the value is always 1.
15:37:52 <dougsz_> (we use log-metrics to generate metrics from specific logs)
15:38:19 <witek> I think it would make sense to have it consistent and include value as well
15:38:44 <witek> now it's only mentioned in the reason string
15:39:12 <witek> which is not nice for making custom notification format
15:39:44 <witek> another question, how are compound alarms handled?
15:39:48 <Charana> yeah i can include that in the notification from the raw alarm
15:40:27 <dougsz_> deferred to Charana
15:41:25 <Charana> https://review.openstack.org/#/c/603127/4/tests/test_alarm_processor.py
15:41:26 <dougsz_> You planned to concatenate value meta from the sub-expressions?
15:41:38 <Charana> if you look at line 149, it would look something like that
15:42:04 <Charana> you could move whatever you felt was necessary from that obj in the alarm to the notification
15:43:48 <witek> why do you inclue `state` and `subexpression` in value_meta?
15:44:20 <witek> `value_meta_obj` I mean
15:45:06 <dougsz_> Some good questions witek
15:45:41 <dougsz_> Shall we continue the discussion on the review?
15:45:44 <Charana> that object is there in the alarm and it helps you determine the state and value_meta of the individual subAlarms when the Alarm was triggered
15:46:15 <witek> yeah, I think it's a good idea to continue the discussion in review
15:46:46 <dougsz_> great thanks witek
15:46:50 <witek> btw. just before the meeting I created Monasca Board for Stein release
15:46:56 <witek> https://storyboard.openstack.org/#!/board/111
15:47:13 <witek> started with adding your story there
15:47:38 <dougsz_> ok, thanks, I'll update it
15:48:03 <Charana> cheers
15:48:13 <witek> #topic Forum in Berlin
15:48:24 <witek> I'd like to jump to that one
15:48:38 <witek> next week is the deadline for proposing Forum sessions
15:49:14 <witek> September 26
15:49:50 <witek> do we have any topics we would like to discuss with users or operators?
15:50:20 <witek> we could also reserve space for our team discussion
15:50:46 <dougsz_> I'm interested in attending
15:51:20 <dougsz_> Operation at scale is of great interest to us
15:52:27 <dougsz_> I'm not sure what sort of deployments people are managing, but it would be interesting to learn more from people operating multi-node cluster and the optimisations that they've made
15:52:38 <dougsz_> To feed back into the Kolla multi-node deployment
15:53:40 <joadavis> I'm trying to sort my thoughts into usable questions.  What features do users feel are missing?  How can we show that the resource use of Monasca is worth the cost?
15:53:59 <joadavis> What do you do with the data (and what do you wish you could do, like billing)?
15:54:28 <witek> I though about asking for feedback on events usage
15:54:35 <joadavis> +1
15:55:20 <witek> do we want to make again a general session with these few questions?
15:56:10 <dougsz_> sounds good to me
15:56:26 <witek> anyone volunteering for moderation?
15:56:54 <witek> I'll set up an etherpad and post a link to the mailing list
15:57:00 <joadavis> I don't have plans to be there in person. :(
15:57:14 <witek> we can collect ideas there
15:57:19 <witek> and decide what to do
15:57:47 <dougsz_> sounds good
15:58:08 <witek> #topic release liaison
15:58:19 <witek> I'm looking for one :)
15:58:35 <witek> if someone would like to do that job, please ping me
15:58:48 <witek> here are the responsibilities:
15:58:54 <witek> https://docs.openstack.org/project-team-guide/release-management.html#release-liaisons
15:59:13 <witek> #topic Pre Upgrade checks
15:59:21 <witek> https://storyboard.openstack.org/#!/story/2003657
15:59:28 <witek> https://storyboard.openstack.org/#!/story/2003657
15:59:34 <witek> https://review.openstack.org/#/c/603465/
16:00:06 <witek> I think they started implementing oslo middleware for this
16:00:08 <witek> https://github.com/cybertron/oslo.upgradecheck
16:00:29 <witek> it might be worth waiting a bit and checking if we can use it
16:00:39 <witek> I have to close now
16:00:50 <joadavis> I posted a simple commit that would do the minimum until oslos sorts out
16:00:56 <dougsz_> ok, thanks all
16:00:58 <joadavis> Thanks all!
16:01:00 <witek> let's move the leftovers to the next week's meeting
16:01:05 <witek> thanks
16:01:06 <witek> bye
16:01:10 <witek> #endmeeting