15:00:56 <rhochmuth> #startmeeting monasca
15:00:57 <openstack> Meeting started Wed Nov 30 15:00:56 2016 UTC and is due to finish in 60 minutes.  The chair is rhochmuth. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:58 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:01:00 <openstack> The meeting name has been set to 'monasca'
15:01:12 <rhochmuth> o/
15:01:19 <rhochmuth> https://etherpad.openstack.org/p/monasca-team-meeting-agenda
15:01:25 <bklei> o/
15:01:28 <koji> o/
15:01:37 <rhochmuth> Agenda for Wednesday November 30, 2016 (15:00 UTC)
15:01:37 <rhochmuth> 1.	Leftovers from last meeting
15:01:37 <rhochmuth> 1.	Agent requirement out-of-sync issue
15:01:37 <rhochmuth> 1.	https://review.openstack.org/#/c/397018/
15:01:37 <rhochmuth> 2.	https://review.openstack.org/#/c/315792/
15:01:37 <rhochmuth> 2.	Use oslo.db for sqla driver (yingjun)
15:01:38 <rhochmuth> 1.	https://review.openstack.org/#/c/395897/
15:01:38 <rhochmuth> 3.	https://storyboard.openstack.org/ - should we use that ?
15:01:39 <qwebirc15108> o/
15:01:39 <rhochmuth> 4.	InfluxDB 0.11 and beyond
15:01:39 <rhochmuth> 2.	Removing pure mysql code ? Is it needed or is it a dead-code ?
15:01:40 <hosanai> o/
15:01:44 <rhochmuth> hi everyone
15:01:45 <shinya_kwbt> o/
15:02:02 <bklei> hello
15:02:17 <rhochmuth> i'm in germany at the open source monitoring conference OSMC 2016
15:02:24 <rhochmuth> giving a talk tomorrow on monasca
15:02:30 <bklei> nice
15:02:39 <rhochmuth> but, i'm here for 30-45 minutes
15:02:53 <rhochmuth> looks like a descent agenda
15:02:58 <rhochmuth> we should get started
15:03:07 <rhochmuth> #topic Agent requirement out-of-sync issue
15:03:16 <rhochmuth> 1.	https://review.openstack.org/#/c/397018/
15:03:16 <rhochmuth> 2.	https://review.openstack.org/#/c/315792/
15:04:43 <rhochmuth> looks like tomasz is not here today
15:04:56 <rhochmuth> and i'm guessing he added this to the agenda
15:05:04 <rhochmuth> does anyone else know what is going on
15:05:18 <rhochmuth> normally, i wait for the openstack bot to handle this sort of stuff
15:05:32 <rhochmuth> but tomasz updated the requirements
15:05:47 <rhochmuth> and the patch is in merge conflict
15:05:54 <rhochmuth> so, can't be merged
15:06:06 <rhochmuth> if no one says anything, we should move on
15:06:25 <rhochmuth> #topic Use oslo.db for sqla driver (yingjun)
15:06:34 <rhochmuth> https://review.openstack.org/#/c/395897/
15:07:25 <rhochmuth> possibly tomasz added this to the agenda too
15:07:49 <rhochmuth> i can't say i'm all to familiar with what the implications are
15:08:41 <bklei> me neither
15:08:45 <rhochmuth> #topic storyboard
15:09:11 <rhochmuth> https://storyboard.openstack.org/ - should we use that ?
15:09:29 <rhochmuth> so, there have been some discussion recently on switching to using storyboard
15:09:35 <rhochmuth> i haven't tried it out
15:09:52 <rhochmuth> but, in general, compare to using launchpad for blueprints it seems like a huge improvement
15:10:12 <rhochmuth> i don't think we should use for bug tracking based on some of the comments so far
15:10:43 <rhochmuth> so, i think we'll need to get back to this topic next week too
15:10:55 <rhochmuth> as whoever posted this to agenda is not here
15:11:11 <rhochmuth> #topic InfluxDB 0.11 and beyond
15:11:35 <rhochmuth> https://review.openstack.org/#/c/396457/
15:11:50 <rhochmuth> this is a brilliant pice of code and should be merged immediately IMHO!
15:12:23 <rhochmuth> from my viewpoint the code is done
15:12:43 <bklei> works with old/new?
15:12:48 <rhochmuth> i've commented on some of the questions related to unit tests, which i don't feel are necessary or even desired in this case
15:12:56 <rhochmuth> yes, it works with old and new
15:13:01 <bklei> sweet
15:13:07 <rhochmuth> that was a lot of work to make sure that was supported
15:13:17 <rhochmuth> more work than the actual code change that is
15:13:41 <bklei> ship it!
15:13:43 <rhochmuth> i've also added comments around my strategy for error handling in the code
15:14:15 <rhochmuth> since i can't really tell waht is an error or expected, it is just json, and there isn't a well defined schema, i test and just continue on
15:14:32 <rhochmuth> if it doesn't match what I'm looking for in the result set from influxdb
15:14:53 <rhochmuth> i didn't want to add lot's of messages to log files
15:15:12 <rhochmuth> anyway, if anyone feels differently, that would be a debate
15:15:28 <rhochmuth> not saying i got it 100% perfect
15:15:50 <rhochmuth> my biggests counter argument woudl be do we test responses from SQL?
15:16:09 <rhochmuth> we don't, other than if it was succesful
15:16:28 <rhochmuth> but, maybe i should add some log messages or just let an exception occur
15:16:39 <rhochmuth> you'll need to look at the code to no what i'm talking about
15:17:00 <rhochmuth> #topic Removing pure mysql code ? Is it needed or is it a dead-code ?
15:17:13 <rhochmuth> i think we shoudl work on removing all the pure mysql code
15:17:22 <rhochmuth> from the python implementation at least
15:17:39 <rhochmuth> i would rather not touch the java implementation at this point
15:18:09 <rhochmuth> we've got a lot of testing on the java mysql code and i don't want to introduce any bugs
15:18:12 <rhochmuth> sorry
15:18:38 <rhochmuth> but, the python implementation cold have the mysql specific code removed
15:18:48 <rhochmuth> at least from an hpe stand-point
15:19:14 <rhochmuth> the notification engine is not that complicated, so risk of introducing a bug is low
15:19:30 <rhochmuth> however, the risk of a bug in the java api and threshold engine, is much higher
15:19:55 <rhochmuth> #topic open floor
15:20:02 <rhochmuth> does anyone have any other topics to discuss?
15:20:06 <bklei> i have one
15:20:11 <rhochmuth> too late
15:20:12 <rhochmuth> bye
15:20:14 <bklei> :)
15:20:16 <rhochmuth> just kidding
15:20:22 <bklei> are we going to meet at https://www.openstack.org/ptg/
15:20:23 <bklei> ?
15:20:32 <bklei> i don't see monasca listed wed-thu
15:20:38 <rhochmuth> i wan't planning on it
15:20:54 <rhochmuth> i don't think we'll get enough of the community to attend
15:21:04 <rhochmuth> so hard to justify IMHO
15:21:19 <bklei> since it's in the US, we at charter are likely to be able to swing it this time
15:21:20 <rhochmuth> do you have a different opinion
15:21:43 <rhochmuth> yeah, i know, but it would be just you guys and me there
15:21:49 <rhochmuth> we can to that in fort collins
15:22:04 <bklei> sure -- that works too
15:22:04 <rhochmuth> i checked with others outside of the usa, and travel will be hard to justify
15:22:12 <bklei> gotcha
15:22:18 <rhochmuth> i think we shoudl to a mid-cycle though
15:22:21 <bklei> well -- if things change, we're likely to attend
15:22:23 <rhochmuth> but a remote one again
15:22:30 <rhochmuth> that is nice to know
15:22:45 <rhochmuth> on a side-note
15:23:07 <rhochmuth> one of the reasons i might want to attend is to drive inter-project initiatives
15:23:18 <rhochmuth> one that i think woudl work well is instrumenting openstack
15:23:30 <rhochmuth> i'e been looking at promethues a lot lately
15:24:09 <rhochmuth> michael hoppal has posted a plugin for the monasca agent that scrapes apps that have been instrumented with the promethieus client libraries, or scrapes promethues exporters
15:24:23 <rhochmuth> there is a library for python
15:24:57 <rhochmuth> so, i'm wondering if instrumenting openstack services with the "promethues" client library would be a good direction adn if openstack would be interesteed in that direction
15:25:08 <rhochmuth> the library is largely independent of promethues
15:25:15 <rhochmuth> it isn't like it is just a promethues thing
15:25:24 <bklei> interesting
15:25:45 <rhochmuth> and as i say with our promethues client plugin, https://review.openstack.org/#/c/401413/, everything would be applicable to monasca too
15:25:48 <jobrs> to be on the safe side we also developed a plugin
15:26:09 <rhochmuth> yes, does your plugin scarpe promethus plugins, or does it scrape the promethues feeration api
15:26:17 <rhochmuth> i thought it was the federation api
15:26:28 <rhochmuth> but did i missunderstand
15:26:41 <rhochmuth> because i didn't want to overlap with your work
15:26:47 <jobrs> that  is right
15:26:50 <jobrs> https://github.com/sapcc/monasca-agent/blob/master/monasca_agent/collector/checks_d/prometheus.py
15:27:12 <rhochmuth> ok, so we aren't over-lapping then
15:27:14 <jobrs> actually we can handle both
15:27:23 <rhochmuth> ohhh, i missed that
15:27:27 <rhochmuth> sorry
15:27:33 <rhochmuth> i hope we are not stepping on your toes
15:27:38 <rhochmuth> that was my last intention
15:27:50 <rhochmuth> i pointed michael at your plugin
15:27:52 <jobrs> no, not at all. let's just look whether we can join forces here
15:28:02 <rhochmuth> absolutely!!!
15:28:31 <jobrs> we have some nice mapping functionality for example
15:28:57 <jobrs> let's have a separate call on that
15:29:18 <rhochmuth> sure, i'm out this week, but next week will work well
15:29:31 <rhochmuth> we'll need to get michael invovled too
15:29:56 <jobrs> sure, you set up a meeting? or just send me this contact details so that I can follow up
15:29:59 <jobrs> thanks
15:30:04 <rhochmuth> jobrs: what do you think about instrumenting openstack with the promethues python client library
15:30:14 <rhochmuth> jobrs: please setup meeting
15:30:19 <rhochmuth> i'm still traveling this week
15:30:23 <jobrs> ok
15:31:01 <jobrs> well, I think statsd is good for push and prometheus might end-up to be the defect standard for pull metrics
15:31:03 <rhochmuth> michael dot jam dot hoppal at hpe dot com
15:31:13 <jobrs> defacto, not defect :-)
15:31:30 <rhochmuth> yes, that is what i'm thinking too
15:31:46 <jobrs> statsd is easier and less risky IMHO
15:32:01 <rhochmuth> but, i think it is a good opportunity for openstack instrumentation to leverage promethus momentum
15:32:06 <jobrs> no open port, no need to buffer in the application, no need to thing about protecting access
15:32:11 <rhochmuth> or to build on that
15:32:30 <rhochmuth> and, i don't see any issues with monasca leveraging that work
15:32:54 <rhochmuth> i think gettign openstack to adopt promethus would be easier than getting them to adopt statsd or monasca-statsd
15:32:59 <rhochmuth> that was my thinking
15:33:05 <rhochmuth> or reasoning
15:33:33 <rhochmuth> and, in the container world, with openstack be run more and more in containers, it would be nice to support both promethues and monasca in one system
15:33:42 <jobrs> it would be great to have some non-intrusive instrumentation tool that keeps instrumentation separate from the code
15:33:45 <rhochmuth> and leverage auto-dscovery
15:34:08 <jobrs> agreed
15:34:16 <rhochmuth> like a layer/wrapper ontop of promethues or monasca-statsd
15:34:23 <rhochmuth> so, you could use either
15:34:47 <jobrs> that would be perfect
15:35:18 <rhochmuth> oslo.instrumentation
15:35:36 <jobrs> I believe statsd+datadog extensions would fit both worlds
15:36:02 <jobrs> there are several lean statsd-exporters for prometheus already
15:36:08 <jobrs> (written in Go)
15:36:40 <rhochmuth> interesting
15:36:44 <rhochmuth> i missed that
15:36:59 <jobrs> having something similar that is pushing directly to Monasca API would defer the decision
15:37:18 <rhochmuth> well, that would be the dvantage of a wrapper library
15:37:19 <jobrs> to those assembling the containers / pods
15:37:23 <rhochmuth> monasca could be directly used
15:38:29 <rhochmuth> i'm going to have to leave in a minute
15:38:50 <rhochmuth> jobrs: i'll meet with you next week along with michael
15:38:58 <jobrs> ok
15:39:08 <rhochmuth> if anyone else want to attend, please let jobrs know
15:39:22 <rhochmuth> this isn't a closed meeting
15:39:35 <rhochmuth> bklei: are you good?
15:39:45 <bklei> yup -- thx roland
15:40:06 <JamesGu_> jobrs and Roland: I do like to attend
15:40:11 <rhochmuth> ok, i can end the meeting or let you guys continue on
15:40:20 <rhochmuth> JamesGu_: OK
15:40:31 <rhochmuth> please let jobrs know and send email
15:40:40 <rhochmuth> thx
15:40:45 <JamesGu_> sure thx
15:40:48 <jobrs> to joachim.barheine@sap.com
15:41:33 <JamesGu_> thx jobrs
15:41:38 <rhochmuth> ok, i've got to run
15:41:45 <rhochmuth> sorry about leaving early
15:42:03 <rhochmuth> i'm going to close meeting in 30 seconds unless i hear otherwise
15:42:53 <rhochmuth> #endmeeting