06:00:26 <dangtrinhnt> #startmeeting Telemetry for Train
06:00:27 <openstack> Meeting started Tue Apr  9 06:00:26 2019 UTC and is due to finish in 60 minutes.  The chair is dangtrinhnt. Information about MeetBot at http://wiki.debian.org/MeetBot.
06:00:28 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
06:00:30 <openstack> The meeting name has been set to 'telemetry_for_train'
06:00:41 <dangtrinhnt> hi all
06:00:47 <mrunge> good morning!
06:00:56 <shaoman> hi
06:00:58 <dangtrinhnt> so we have hyang, zhurong_phone, shaoman
06:00:59 <zhurong_phone> hi all
06:01:07 <hyang> hi there!
06:01:07 <dangtrinhnt> joadavis, are you there?
06:01:30 <dangtrinhnt> witek hi
06:01:39 <dangtrinhnt> mrunge cannot join
06:01:46 <dangtrinhnt> lxkong is not here I guess
06:01:49 <witek> hi dangtrinhnt
06:01:52 <mrunge> https://etherpad.openstack.org/p/telemetry-train
06:01:59 <mrunge> dangtrinhnt: sorry?
06:02:02 <dangtrinhnt> ah
06:02:08 <dangtrinhnt> you are here, mrunge
06:02:15 <mrunge> yes, I am
06:02:19 <dangtrinhnt> ok, great
06:02:55 <dangtrinhnt> #topic Active contributors/cores
06:03:11 <dangtrinhnt> anyone of you is the current core of Telemetry?
06:03:24 <dangtrinhnt> zhurong_phone?
06:03:55 <dangtrinhnt> mrunge?
06:04:09 <mrunge> dangtrinhnt: I am not, but I know who to ping
06:04:27 <mrunge> dangtrinhnt: the list is in gerrit
06:04:28 <zhurong_phone> nup, I know 2 guys from China are cores, but they leaved
06:04:30 <witek> https://review.openstack.org/#/admin/groups/107,members
06:04:40 <mrunge> thank you witek
06:04:47 <dangtrinhnt> Thanks. I'm expecting some existing core reviewers to join but seems like everyone gone.
06:05:02 <mrunge> or in US timezone
06:05:21 <zhurong_phone> almost gone
06:05:34 <dangtrinhnt> ok, no worries. It means that we have to setup a new team.
06:06:07 * mrunge taking notes
06:06:21 <dangtrinhnt> So, I assume everyone here will be active contributors
06:06:25 <mrunge> -> etherpad, linked above
06:06:36 <dangtrinhnt> Tanks mrunge.
06:06:37 <dangtrinhnt> thanks
06:07:04 <witek> dangtrinhnt: that's an optimistic approach :)
06:07:16 <mrunge> dangtrinhnt: yes!
06:07:57 <dangtrinhnt> so, mrunge, witek, zhurong_phone, shaoman, lxkong, hyang I guess you guys somewhat have experience with Ceilometer/AODH
06:08:37 <mrunge> dangtrinhnt: somewhat, yes
06:08:45 <witek> I would add joadavis to the list
06:08:56 <dangtrinhnt> ah, my bad, joadavis too
06:09:01 <hyang> yes, for me we have Ceilometer/Aodh/Gnocchi/Senlin running in production so I have experience in using but not much about contributing
06:09:16 <dangtrinhnt> hyang, that's awesome as well.
06:10:15 <zhurong_phone> yep, we are heavily user
06:10:47 <dangtrinhnt> ok
06:11:00 <shaoman> We now use ceilometer/aodh/panko
06:12:12 <mrunge> so, interested folks are mostly users, no?
06:12:50 <dangtrinhnt> Ok, It's good. After the meeting, I will try to update the plan for Train and we can officially update the cores. In the mean time, please share your experience with Telemetry via mailing list.
06:13:20 <hyang> sounds good!
06:13:56 <dangtrinhnt> I will make the process fast, 1 week maybe. To me, the criteria is experience with Telemetry projects and willing to contribute.
06:14:13 <dangtrinhnt> Please raise your hand if you have any objection for that.
06:14:24 <zhurong_phone> I maybe don't have many time on telemetry, but shaoman will be.
06:14:33 <dangtrinhnt> Thanks mrunge for taking notes
06:14:38 <zhurong_phone> agree
06:14:57 <mrunge> I would argue, having experience does not necessarily qualify for code contributions, but let's see....
06:15:31 <dangtrinhnt> mrunge, make senses, that I think will need more discussion.
06:16:12 <mrunge> dangtrinhnt: you are the only core for now left?
06:16:40 <dangtrinhnt> I'm just added
06:16:56 <dangtrinhnt> unfortunately
06:17:03 <mrunge> and nobody really has code contribution experience?
06:17:20 <dangtrinhnt> with other projects but not with telemetry
06:17:28 <mrunge> yes
06:17:28 <dangtrinhnt> That's the problem we're trying to solve.
06:17:52 <zhurong_phone> mrunge: I think we need more contributors, we can do it like the initial repo, first add the actives, and then quickly update
06:18:14 <mrunge> uhm, no. you're trying to solve something difference, dangtrinhnt
06:18:23 <mrunge> dangtrinhnt: you're trying to keep projects alive
06:18:27 <mrunge> that is different
06:18:36 <dangtrinhnt> mrunge. Ah yes.
06:19:15 <mrunge> dangtrinhnt: ok, since you are the only core left (for now), mergeing a patch will only require one core +2?
06:19:22 <mrunge> how about other reviews?
06:19:39 <mrunge> .... this creates "rules" for merging code
06:20:05 <dangtrinhnt> we need more reviewers of-course. I'm trying to find other experience developers.
06:20:20 <dangtrinhnt> lxkong and people from zhurong_phone team may help
06:20:53 <mrunge> we may have some time to review patches
06:21:09 <mrunge> and also to put up patches
06:21:10 <dangtrinhnt> Also, I would trying to communicate with the previous cores if I can
06:21:21 <dangtrinhnt> sure
06:23:15 <dangtrinhnt> mrunge, I don't think it is a problem since, I believe, all of us have experience in contributing to openstack projects.
06:23:55 <mrunge> dangtrinhnt: let's get started and fix processes when we see issues
06:24:04 <dangtrinhnt> agree
06:24:08 <dangtrinhnt> ok, anything else?
06:24:26 <mrunge> question: what does it take to get a patch merged?
06:24:52 <mrunge> usual openstack projects: 2 cores +2/A
06:25:00 <mrunge> with only one core left...?
06:25:16 <dangtrinhnt> I will ask for more cores after this.
06:25:25 <mrunge> ack
06:26:03 <mrunge> next topic: Current state?..
06:26:04 <zhurong_phone> I think we need more cores and more fresh man
06:26:36 <dangtrinhnt> to me, for now, the process could be: 1. Pass the CI, 2. The feature aligns with Telemetry roadmap, 3. well tested,
06:26:51 <dangtrinhnt> ok, let's move on the next topic
06:27:08 <dangtrinhnt> #topic Current state of Telemetry
06:28:27 <dangtrinhnt> So, now we have ceilometer, aodh, panko under OpenStack umbrella and seems like most stuff were moved to gnocchi
06:29:10 <zhurong_phone> yes, exactly
06:29:16 <mrunge> yes
06:30:36 <dangtrinhnt> like the other day talk to lxkong and someone about cpu_utils has been removed and we have to calculate it in gnocchi
06:31:07 <dangtrinhnt> What do you think about that? what should we move forward with that?
06:31:09 <sapd1> yep
06:31:27 <dangtrinhnt> sapd1 hi
06:31:34 <hyang> afaik, aodh and panko are in maintenance mode currently?
06:31:54 <mrunge> yes, that's my understanding, hyang
06:31:58 <dangtrinhnt> I think so.
06:32:26 <mrunge> dangtrinhnt: what is the question, I did'nt really get you
06:32:28 <zhurong_phone> afaik panko has deprecated in RHOP
06:32:35 <mrunge> move forward with what?
06:33:01 <dangtrinhnt> with the way telemetry has to depend on gnocchi, an external project.
06:33:32 <zhurong_phone> I think we can't rollback
06:33:34 <mrunge> yes, that is my understanding
06:33:39 <mrunge> what is the alternative?
06:33:47 <witek> we could think about alternative backends, like Monasca
06:34:09 <mrunge> is monasca a backend?
06:34:20 <hyang> dangtrinhnt: yes, I asked a similar question about cpu_util last week and lxkong replied me that he also got the same issue earlier. Currently after Ceilometer deprecated the transformer support, there is no clear way to get the cpu_util metric
06:34:22 <witek> it can be, yes
06:34:29 <mrunge> ... isn't the data store used by monasca the backend?
06:34:37 <dangtrinhnt> mrunge, gnocchi is now a big problem too because Julien changed his jobs and not sure about its health
06:34:59 <mrunge> dangtrinhnt: I agree, yes.
06:35:20 <mrunge> dangtrinhnt: one alternative might be to move gnocchi back under openstack umbrella
06:35:23 <dangtrinhnt> witek, I believe monasca use InfluxDB as storage backend
06:35:25 <shaoman> yes, since ceilometer storage(mongodb) and api is removed, gnocchi must be used, and more things moved from ceilometer to gnocchi
06:35:33 <mrunge> dangtrinhnt: yes, that's right
06:35:54 <witek> dangtrinhnt: mrunge: Monasca officially supports InfluxDB and Cassandra as TSDB
06:36:15 <mrunge> one possible alternative could be to store metrics in prometheus
06:37:02 <dangtrinhnt> ok, mrunge, witek, let's put them into a list and we will select which one will be the first to replace gnocchi if needed.
06:37:05 <mrunge> how many contributors has monasca?
06:37:11 <dangtrinhnt> not sure about that
06:37:21 <mrunge> ... don't want to add another understaffed project as dependency
06:37:28 <dangtrinhnt> agree
06:37:32 <zhurong_phone> agree
06:37:38 <dangtrinhnt> InflexDB seems like a strong project
06:37:38 <witek> we have an active team of around 10 contributors
06:37:54 <dangtrinhnt> prometheus is great too
06:38:02 <sapd1> If we use prometheus, we can use AlertManager too.
06:38:06 <mrunge> oh, I didn't knew about that witek, that is great to hear
06:38:11 <mrunge> yes, sapd1
06:38:16 <hyang> I think both Monasca and prometheus are pretty active
06:38:16 <sapd1> instead of aodh.
06:38:41 <mrunge> there are gaps with prometheus, but let's see if that really hurts us
06:38:54 <dangtrinhnt> so, what you're saying is we can possibly remove AODH?
06:39:00 <dangtrinhnt> mrunge, sure.
06:39:22 <mrunge> yes, that's right
06:39:36 <mrunge> ... shrinking down would be great, imho
06:39:44 <dangtrinhnt> Agree. I would suggest putting prometheus support as a feature for Train.
06:39:47 <mrunge> ... re-scoping
06:39:53 <dangtrinhnt> sure
06:40:20 <sapd1> Oops. So because multi-tenancy feature, We can't remove aodh.
06:41:15 <dangtrinhnt> ok, I would suggest talking about it in the team meeting.
06:41:20 <dangtrinhnt> let's move on to the next topic
06:41:27 <zhurong_phone> this is a really big change, I think we need a maillist to discuss more
06:41:37 <shaoman> a lot of work has been done for different publishers, but do we need a offically backend and api?
06:41:38 <dangtrinhnt> ok, zhurong_phone
06:42:01 <hyang> So maybe I miss something, but what's the difference between Ceilometer and Monasca-agent? If we keeps only Ceilometer in the Telemetry project and integrate with Monasca, why not just Monasca alone?
06:42:22 <dangtrinhnt> hyang, that's a great question.
06:42:49 <shaoman> yes, agree
06:42:51 <joadavis> There are a few differences between ceilometer agent and monasca agent
06:42:54 <witek> monasca-agent does not collect information from OpenStack notifications
06:42:55 <joadavis> (and hi)
06:43:08 <dangtrinhnt> It's a big question actually, let's talk about it in  the mailing list.
06:43:27 <dangtrinhnt> we only have 17m left
06:43:29 <hyang> sure, I'm ok to move on
06:43:32 <dangtrinhnt> let's move to the next topic
06:43:42 <dangtrinhnt> #topic Existing bugs
06:44:19 <dangtrinhnt> someone put this: ceilometer CI gate telemetry-dsvm-integration seems to be broken for older branches
06:45:13 <joadavis> I've noticed that over the last few commits
06:45:47 <dangtrinhnt> I'm trying to fix that. Any ideas? anyone?
06:45:49 <joadavis> I don't have enough knowledge to know what the hangup is, though I suspect older branches haven't been updated to match current Zuul requirements
06:46:05 <mrunge> e.g this one: https://review.openstack.org/#/c/646055/
06:46:13 <mrunge> that was a recent change
06:46:38 <dangtrinhnt> joadavis, it makes sense.
06:46:43 <mrunge> the error was something with heat
06:47:19 <dangtrinhnt> ok, I will talk to the Heat and Infra team for help.
06:47:31 <joadavis> another example - https://review.openstack.org/#/c/640223/ (which incidentally may need discussion before merging)
06:47:57 <zhurong_phone> dangtrinhnt: I and shaoman will take a look too, our teme is heat/ceilometer
06:48:10 <dangtrinhnt> thanks, zhurong_phone
06:48:17 <mrunge> awesome, thank you zhurong_phone
06:48:54 <dangtrinhnt> ok, let's move on
06:49:03 <dangtrinhnt> #topic TODOs for Train
06:49:12 <dangtrinhnt> 10 more minutes left
06:50:13 <mrunge> it seems we need much more discussion
06:50:30 <dangtrinhnt> yes
06:50:38 <witek> is anyone attending the Summit and PTG in Denver?
06:50:47 <mrunge> nope
06:50:49 <dangtrinhnt> I would suggest we focus on setting up the team and the roadmap
06:50:54 <joadavis> I will be there for Monasca related items
06:51:05 <joadavis> +1 for roadmap
06:51:08 <hyang> yes, I'll be in the PTG
06:51:10 <dangtrinhnt> witek, unfortunately, my visa application has been rejected
06:51:32 <zhurong_phone> I am appliying visa
06:51:34 <hyang> dangtrinhnt: that's so sad to hear...
06:51:49 <joadavis> I'm sorry you won't be there
06:52:11 <dangtrinhnt> ok, let have a quick look at this #link https://wiki.openstack.org/wiki/Telemetry/RoadMap
06:52:51 <joadavis> I'd asked jd about the roadmap not too long ago, and he said there isn't one (I can dig that up in IRC logs if needed)
06:53:06 <dangtrinhnt> we need to update it
06:54:28 <mrunge> maybe collecting items in etherpad first?
06:54:37 <mrunge> and then see what we can do in stein?
06:54:38 <dangtrinhnt> mrunge, sure.
06:54:57 <dangtrinhnt> Train you mean?
06:55:05 <mrunge> yes, sure!
06:55:14 <mrunge> and beyond
06:55:26 * mrunge needs a coffee, apparently
06:55:36 <dangtrinhnt> #link https://etherpad.openstack.org/p/telemetry-train-roadmap
06:56:25 <dangtrinhnt> please put your ideas to the telemetry roadmap and we will discuss over mailing list
06:56:35 <joadavis> a clear statement of intended use and architectures would be helpful too.  That may come out of defining the roadmap
06:56:49 <dangtrinhnt> something like a vision for Telemetry. +1 joadavis
06:57:06 <mrunge> yes, that's a good idea
06:57:23 <dangtrinhnt> ok, 3m left. Anything else?
06:57:35 <mrunge> next meeting?
06:57:42 <mrunge> ... in one week?
06:57:53 <dangtrinhnt> ok, weekly I think is best
06:58:00 <mrunge> or can we skip next week?
06:58:09 <dangtrinhnt> Will send out and email asking for update the meeting time
06:58:28 <mrunge> I'll be on vacation, and can't guarantee on network connectivity
06:59:18 <witek> same for me
06:59:33 <dangtrinhnt> ok, no problem. We can figure that out, maybe in 2-3 weeks.
06:59:47 <dangtrinhnt> ok, time up
07:00:07 <mrunge> thank you dangtrinhnt  for organizing this meeting
07:00:09 <dangtrinhnt> Thanks everyone for joining! Let's make Telemetry great again! :)
07:00:18 <dangtrinhnt> #endmeeting