14:02:02 <rhochmuth> #startmeeting monasca
14:02:03 <openstack> Meeting started Wed Aug  9 14:02:02 2017 UTC and is due to finish in 60 minutes.  The chair is rhochmuth. Information about MeetBot at http://wiki.debian.org/MeetBot.
14:02:04 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
14:02:06 <openstack> The meeting name has been set to 'monasca'
14:02:09 <kornica> o/
14:02:19 <witek> hello
14:02:24 <joadavis> hiya
14:02:32 <jgu> Hello
14:02:33 <haruki> o/
14:02:35 <hosanai> o/
14:03:19 <koji> o/
14:03:28 <rhochmuth> so no one has posted any agenda items
14:03:54 <Fdaisuke> o/
14:03:58 <rhochmuth> are there any topics to discuss
14:04:14 <jgu> is there a Monasca ptg?
14:04:30 <jgu> didn't see on the openstack schedule
14:04:43 <rhochmuth> we don't have a monasca ptg planned
14:05:01 <rhochmuth> there weren't a critical mass of folks that could get travel approval
14:05:15 <jgu> okay recall Witek was asking at one time, so wasn't sure
14:05:22 <rhochmuth> however, there is a mid-cycle that is possible
14:05:29 <witek> but we should organize mid-cycle meeting remotely
14:05:32 <rhochmuth> via remote conferencing
14:05:46 <jgu> sounds good
14:06:11 <rhochmuth> are you planning on attending the ptg anyway?
14:06:12 <witek> any ideas when?
14:06:25 <jgu> if you an witek will be there, sure
14:07:06 <witek> I don't think I will get approval
14:07:27 <witek> but some things changed, so perhaps
14:08:05 <jgu> here we just got approval :-)
14:08:47 <witek> to be honest, I didn't plan to go to Denver
14:08:57 <witek> also, Monasca did not book the room for sessions
14:09:07 <rhochmuth> correct
14:09:15 <witek> because there was no interest/need
14:09:39 <witek> I think it might be too late to organize it now
14:09:42 <kornica> that sounds sad :(
14:10:31 <witek> so I would suggest to start planning the remote mid-cycle meeting
14:10:36 <jgu> :-). when will be the mid cycle remote conferencing?
14:10:57 <witek> we could do this one week after PTG
14:11:47 <jgu> sounds good to me
14:11:48 <witek> preferred times?
14:12:21 <witek> last time we needed two days
14:12:43 <witek> 4 hours on each day
14:13:00 <witek> if I recall correctly
14:13:37 <jgu> yes, I remember we started at 7am PST.
14:14:08 <witek> is it OK for everyone?
14:14:46 <rhochmuth> i think tht will work for me
14:15:05 <jgu> works for me too.
14:15:15 <rhochmuth> so, what would the topics be?
14:15:44 <witek> I can think of TSDB
14:15:56 <rhochmuth> thats a good one
14:16:03 <jgu> yes
14:16:04 <witek> Docker/Helm
14:16:12 <rhochmuth> ok
14:16:57 <kornica> Events ?
14:17:03 <jgu> retention mgmt?
14:19:36 <witek> there is new monasca-aggregator project similar in goals to monasca-transform
14:19:55 <joadavis> yeah, I'm curious about that ;)
14:19:56 <witek> should we  discuss it?
14:20:59 <witek> rhochmuth: ^^ you're the author
14:21:17 <rhochmuth> yes, i'm one of the authors
14:21:40 <rhochmuth> i think the landing page describes most of the motification for monasca-aggregator
14:22:12 <kornica> I'd like to know which one is the one monasca should go with...seems like both are about the same thing, aren't they ? will the be any re-eval process for it ?
14:22:53 <rhochmuth> monasca-aggregator
14:23:29 <joadavis> why?
14:24:10 <witek> ok, I guess it's a hot topic for mid-cycle :)
14:24:25 <kornica> that and if that's the case should monasca-transform be retired as in https://docs.openstack.org/infra/manual/drivers.html#retiring-a-project
14:24:26 <joadavis> :)
14:24:41 <rhochmuth> my high-level summary is monasca-transform has high cpu utilization, memory utilization, has a much more difficult deployment, due to spark
14:25:15 <rhochmuth> in addition, it is more difficult to add new aggregations, monasca-aggregator has a nice DSL for configuring/modifying aggregation functions
14:26:09 <joadavis> I'd be curious about the overall monasca services roadmap - mon-trans/mon-aggregat, mon-ceilometer, mon-events...
14:26:09 <rhochmuth> i think the landing page, https://github.com/monasca/monasca-aggregator, summarizes the main reasons
14:26:54 <jgu> and it's written in go :-)
14:27:33 <kornica> that's an advantage or disadvantage, depends on POV ;-) :D
14:27:34 <joadavis> well, that seems like a hinderance to me, as go isn't openstack supported.  But I've invested some time in mon-trans, so I will take a bit of convincing. ;)
14:28:18 <kornica> joadavis: well that's because I am asking - there's been some activity recently around mon-transform over review.openstack.org
14:28:45 <kornica> and it kind of odd to have both at the same time
14:28:56 <kornica> I acknowledge abvious pros of the monasca-aggregator
14:29:22 <joadavis> :) I work with Ashwin, and we are looking at continuing work on mon-transform
14:29:22 <kornica> but if that's the way to go, as PTL says so, the monasca-transform is a candidate to be retired according to the Openstack guidelines
14:29:40 <rhochmuth> joadavis: well, we didn't write monasca-aggregator because we had nothing better to do
14:30:00 <rhochmuth> as usual the discussion in openstack degraded to language comparisons goland vs python for example, and prove it
14:30:51 <rhochmuth> sure, you should evaluate monasca-aggregator in that to decide whether monasca-transform is the right direction
14:31:03 <witek> ok, I think I will set up etherpad for mid-cycle agenda and let you decide if you want to discuss it
14:31:25 <witek> which days should we plan?
14:31:29 <rhochmuth> i know full well of the past performance, utilization and complexity of monasca-transform
14:32:47 <jgu> any day of that week is good for me
14:32:53 <rhochmuth> joadavis: speaking about convincing, i
14:33:08 <rhochmuth> i'm not very convinced y the cassandra analysis you have put together
14:33:37 <joadavis> rhockmuth: I'm sure you do, as one of the developers.  We are still working on Ocata era stuff at the moment, so continuing with what we have makes sense.  But if the overall monasca plan goes another way, we want to keep moving forward.
14:33:38 <jgu> rhochmuth: yes I would like to discuss that :-)
14:33:56 <rhochmuth> sorry, meant jgu
14:34:10 <rhochmuth> ok, we can discuss that at the mid-cycle too
14:34:15 <joadavis> 'sallright, I figured that out. ;)
14:35:50 <jgu> rhochmuth: can we start the Cassandra discussion earlier if you have feedback already? we have internal project schedule need to decide the tsdb earlier that September.
14:36:19 <jgu> sorry... *earlier than september*
14:37:15 <rhochmuth> i'll defer to witek on that question
14:37:35 <rhochmuth> i think this would take a special session
14:37:43 <rhochmuth> like a one hour meeting at least
14:38:04 <witek> jgu: I have sent you my questions to the document just today
14:38:53 <jgu> witek: which email account?
14:39:06 <witek> I just answered yours
14:39:27 <jgu> ah got it
14:40:05 <witek> I think it would be good to move the discussion to openstack-dev mailing list, what do you think?
14:41:16 <jgu> that sounds good to me to start it on openstack mailing list.
14:41:21 <rhochmuth> sounds fine to me
14:43:36 <witek> what bothers me somehow is that you plan to go with Java, which we were about to deprecate
14:44:04 <witek> I mean monasca-api and monasca-persister in Java
14:45:18 <jgu> we started with Java is because that's what is being used today.
14:46:02 <jgu> I'd prefer to write in go for the persister :-).
14:46:09 <witek> not because of performance?
14:46:45 <jgu> yes performance -- is that why the persister is in Java today?
14:46:46 <witek> it's not about the language, it's about doubling maintenance efforts
14:47:42 <jgu> the Cassandra driver and client are pretty heavy on multi-threading.
14:51:46 <jgu> we do intend to have a python implementation but wasn't thinking to have python as the first and only language implementation. the previous investigation into Cassandra also characterizes the python Cassandra client throughput at about 1/4 TO 1/3 OF java's?
14:52:53 <witek> or even less
14:53:35 <jgu> right :-)
14:54:53 <witek> we haven't set the dates for mid-cycle, I will create doodle and send to openstack-dev
14:55:20 <witek> I will also create etherpad to start planning the agenda
14:55:26 <jgu> but if aggregator is going to introduce golang, maybe we should reconsider the language choice?
14:55:54 <rhochmuth> the aggregator is not part of openstack code base
14:56:35 <jgu> ah... different than the transform?
14:56:51 <rhochmuth> correct
14:57:19 <rhochmuth> note, golang has been getting some adoption in openstack
14:57:36 <rhochmuth> not sure if that changes anything
14:58:04 <witek> yes, but introducing 3rd lang to persister isn't probably wise
14:58:12 <jgu> yes an interesting (in a good way) change
14:59:38 <witek> before we close the meeting
15:00:01 <witek> I wanted to announce my candidacy for the Project Team Lead
15:00:11 <witek> https://git.openstack.org/cgit/openstack/election/plain/candidates/queens/Monasca/witek.txt
15:01:15 <rhochmuth> thx witek
15:01:20 <rhochmuth> need to end meeting
15:01:27 <rhochmuth> please vote for witek
15:01:30 <rhochmuth> #endmeeting