15:00:18 <rhochmuth> #startmeeting monasca
15:00:18 <openstack> Meeting started Wed Sep  2 15:00:18 2015 UTC and is due to finish in 60 minutes.  The chair is rhochmuth. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:19 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:21 <openstack> The meeting name has been set to 'monasca'
15:00:32 <fabiog> o/
15:00:36 <jimbaker> o/
15:00:42 <rhochmuth> Roll call
15:00:44 <rhochmuth> 0/
15:00:44 <ddieterly> o?
15:00:44 <cdent> o/
15:00:46 <rhochmuth> o/
15:00:46 <rbak> o/
15:00:48 <ddieterly> o/
15:01:01 <bklei_> o/
15:01:03 <witek> o/
15:01:19 <Martin____> o/
15:01:19 <rhochmuth> don't follow me, my typing skills are bad
15:01:25 <rhochmuth> hi everyone
15:01:31 <bklei_> good morning
15:01:33 <rhochmuth> welcome to irc
15:01:42 <bmotz> o/
15:01:49 <rhochmuth> i've added some agenda items at https://etherpad.openstack.org/p/monasca-team-meeting-agenda
15:02:09 <rhochmuth> please add, but we'll also have an open discussion too
15:02:14 <bklei_> bueno
15:02:33 <rhochmuth> so first topic is that fujitsu code
15:02:41 <rhochmuth> congratulation fujitsu
15:02:46 <bklei_> +1
15:02:47 <witek> thank you! :)
15:02:51 <Martin____> thanks :)
15:02:56 <rhochmuth> Deklan got that all tested and merged yesterday
15:03:01 <jimbaker> rhochmuth, sounds good. maybe we can talk briefly about blueflood, our opensourced time series cassandra integration at rackspace, but alex scammon is on vacation this week
15:03:03 <Martin____> that's great news
15:03:09 <fabiog> Martin____: great work!
15:03:14 <rhochmuth> ok jim
15:03:28 <rhochmuth> note, there was a bug that craig found with timezone conversion
15:03:31 <ddieterly> yea, the caveat is that there was some bugs in the code concerning time zones and dates
15:03:40 <Martin____> @fabiog actually I didn't do a single line of coding :)
15:03:43 <witek> we will investigate that
15:03:58 <rhochmuth> ok, let us know if you need more info
15:04:07 <rhochmuth> i'm not sure if  bug was opened in launchpad
15:04:10 <bklei_> question on fujitsu merge -- should i wait to pull in for timezone fix, or are those bugs only logging related
15:04:12 <rhochmuth> but we should do that
15:04:32 <rhochmuth> i'm not sure of the details.
15:04:40 <rhochmuth> deklan do you have more info
15:04:45 <ddieterly> @bklei_ only hibernate related; should not impact twc
15:05:09 <rhochmuth> correct, sorry brad, misunderstood the question
15:05:14 <bklei_> so only related to new code, i'm itching to pull in the new schema and get that out of the way
15:05:29 <ddieterly> yea, only new code path
15:05:34 <rhochmuth> yes, you should test first, and look at the migrations that were done
15:05:42 <bklei_> so question -- new monasca-api, persister and ?
15:05:54 <rhochmuth> threshold engine and monasca common
15:06:03 <ddieterly> not persister; just thresh and api/common
15:06:17 <bklei_> ok perfect
15:06:20 <rhochmuth> yes, don't forget monasca-common
15:06:26 <bklei_> will do
15:06:39 <rhochmuth> note, the nofication engine, Python code, was done a while ago
15:06:46 <bklei_> ok
15:06:52 <rhochmuth> i'm guessing you've already picked that up
15:07:03 <bklei_> not for about a month
15:07:05 <bklei_> or more
15:07:10 <rhochmuth> everything should be compatible, so no config changes should be required
15:07:21 <bklei_> cool
15:07:24 <ddieterly> *should*
15:07:28 <bklei_> will pull in ^^ this week and test
15:07:38 <rhochmuth> thx
15:07:47 <rhochmuth> are we done with that topic
15:07:52 <bklei_> are for me
15:07:57 <ddieterly> sure
15:08:04 <witek> for me too, thanks
15:08:11 <Martin____> ok.. Brad, just let us know if you encounter any problems
15:08:22 <bklei_> will do, thx for the migration path
15:08:26 <rhochmuth> ok, i think there is a way to do topics in irc, but i don't know it yet
15:08:43 <rhochmuth> let's talk about performance at twc
15:08:46 <bmotz> sorry - bit late to this, but which projects has the Fujitsu code gone into
15:08:49 <bmotz> ?
15:08:55 <rhochmuth> brad, what would you like talk about
15:09:12 <rhochmuth> bmotz, monasca-api, thresh, common
15:09:24 <ddieterly> bmotz: monasca-api, monasca-common, ansible-monasca-schema, monasca-thres
15:09:44 <ddieterly> it is all java code with compatible schema changes
15:09:57 <rhochmuth> bklei_ what would you like to do to address performance in terms of planning
15:09:58 <bklei_> my perf issue is a straightforward use case, just display the default libvirt metrics in a graph, on a project with more than a couple instances, unusable
15:10:08 <bmotz> thanks
15:10:28 <rhochmuth> by couple, do you mean 2, or 20
15:10:33 <bklei_> i'm out of 'low hanging fruit' ideas, would like to do a focused whiteboard session, demonstrate the use case
15:10:48 <ddieterly> can u just set up a meeting?
15:11:12 <rhochmuth> besides deklan and me does anyone else want to attend?
15:11:23 <bklei_> sure, can do desktop share meeting
15:11:53 <rhochmuth> if it is just deklan and me, we could to a physical meeting, but if others want to be invovled, can do desktop sharing
15:12:17 <bklei_> ok -- if others want to join, ping me. what day is good for roland and deklan?
15:12:23 <Martin____> NICK mroderus
15:12:42 <deklan> this friday
15:12:43 <bklei_> got it Martin____
15:12:51 <bklei_> friday works
15:12:52 <rhochmuth> firday
15:13:22 <rhochmuth> after 11, hp or twc
15:13:29 <rbak> bklei_: I wouldn't mind being part of that meeting, but I could always come down for that if you want to have the meeting in person.
15:13:50 <rhochmuth> ok, let's take this off-line then
15:13:59 <rhochmuth> and discuss separately
15:14:06 <bklei_> in person would be better, will do
15:14:19 <fabiog> rhochmuth: I am interested
15:14:33 <fabiog> rhochmuth: please
15:14:35 <bklei_> ok - can share desktop for fabiog if we meet physically
15:14:42 <rhochmuth> ok, sounds good
15:15:02 <rhochmuth> time for next topic?
15:15:08 <deklan> yea
15:15:10 <bklei_> good for me
15:15:25 <rhochmuth> Tokyo sessions:
15:15:39 <rhochmuth> Congrats to Fabio and Martin!
15:15:47 <deklan> +1
15:15:50 <bklei_> +1 for roland too
15:15:53 <jimbaker> nice
15:15:55 <mroderus> thanks! looks like Monasca is well represented this time
15:16:03 <deklan> they're just in it for the sushi
15:16:14 <bklei_> and sake
15:16:14 <witek> # /topic Update on Tokyo Summit sessions that were accepted and preparation.
15:16:19 <rhochmuth> yes, there is quite a bit of momentum
15:16:23 <fabiog> yes, we have a good representation
15:16:32 <rhochmuth> three session in total
15:16:46 <rhochmuth> martin, you and i should start preparing
15:16:51 <bklei_> wish i was going, will watch online
15:16:57 <rhochmuth> i've got some slides in my head,
15:17:26 <rhochmuth> maybe get together to discuss soon off-line
15:17:38 <jimbaker> some chance i will go. but otherwise look forward to seeing online
15:17:39 <mroderus> yes, let's have that discussion off-line
15:18:08 <rhochmuth> ok, sounds good
15:18:09 <mroderus> Monasca is also mentioned in two other sessions
15:18:19 <rhochmuth> what sessions were those
15:18:41 <mroderus> http://sched.co/49vh
15:18:54 <mroderus> http://sched.co/49tn
15:19:14 <rhochmuth> Cool, looks like a TWC session
15:19:23 <bklei_> adam's been pushing swift data into monasca for a swift service dashboard here at TWC, very cool
15:19:54 <rhochmuth> the auto-scaling one i believe is from hp
15:20:10 <bklei_> nice, sweet feature
15:20:30 <mroderus> yes, Matt Young. That sessions also sounds interesting
15:20:47 <rhochmuth> ok, next tpoic?
15:20:55 <rhochmuth> governance application
15:21:11 <rhochmuth> there have been some recent updates
15:21:18 <deklan> can we ask that people please +1 the submission if they already have not?
15:21:28 <witek> http://sched.co/49uq
15:21:38 <rhochmuth> see, https://review.openstack.org/#/c/213183/
15:22:01 <rhochmuth> they are doing a roll call, and it is up for discussion at the next governance meeting i beleive
15:22:06 <rhochmuth> i'll need to check the agenda
15:22:31 <rhochmuth> if anyone has comments to add, i think that is ok too
15:23:00 <bklei_> good responses thus far roland
15:23:15 <bklei_> for reading after this meeting http://www.eweek.com/cloud/openstack-moves-from-integrated-release-to-big-tent-model.html
15:23:18 <rhochmuth> so, i plan to address Hay's other comments with an update today or tomorrow
15:23:40 <rhochmuth> we have an eweek article?
15:23:54 <rhochmuth> no way
15:24:06 <bklei_> not mentioned explicitly, but definitely relates to our current submission
15:24:07 <rhochmuth> do you guys want my autograph
15:24:20 <bklei_> signed headshot
15:24:33 <rhochmuth> tattoo
15:25:35 <rhochmuth> ok, unless anyone has further questions on governance, then let's move on
15:25:45 <bklei_> thx for keeping that moving
15:25:53 <rhochmuth> welcome
15:25:57 <mroderus> yeah, great work Roland!
15:26:20 <rhochmuth> next topic is collaboration with Ceilometer
15:26:33 <rhochmuth> there are two promosing areas right right now
15:26:52 <rhochmuth> alarm engine (Aodh in Ceilometer land) and data collection componetization
15:27:04 <jimbaker> rhochmuth, are they planning to rewrite aodh?
15:27:18 <jimbaker> in terms of a new streaming engine?
15:27:22 <rhochmuth> i think it is a start
15:27:32 <rhochmuth> yes, they want to support streaming
15:27:48 <rhochmuth> so, i review Aodh last week
15:28:12 <rhochmuth> in general, for a pure python implementation, it seems like a good start
15:28:12 <fabiog> rhochmuth: Gordon mentioned that NEC is pushing for this solution
15:28:20 <rhochmuth> but not streaming.
15:28:25 <rhochmuth> yes, NEC is very interested
15:28:29 <jimbaker> i went through the existing codebase last week and it does seem to have some hard coded dependencies
15:28:35 <rhochmuth> we need to meet with them
15:28:46 <rhochmuth> correct, jimbaker
15:28:46 <fabiog> rhochmuth: did NEC already submitted a patch somewhere? cdent ?
15:28:59 <cdent> there's is event based alarming pending
15:29:00 <jimbaker> we need to move away from an eventlet model at the very least
15:29:01 <rhochmuth> fabiog, i'm not aware of anything
15:29:02 * cdent locates the review
15:29:23 <rhochmuth> so, jimbaker, you are interested in this topic?
15:29:40 <jimbaker> rhochmuth, yeah, sorry i saw the email, but i didn't have a chance to respond
15:29:45 <jimbaker> but definitely
15:29:54 <cdent> here's one of them: https://review.openstack.org/#/c/200974/
15:30:04 <fabiog> cdent: thanks
15:30:19 <rhochmuth> cdent, ok, i'll review off-line, if that's ok
15:30:23 <rhochmuth> thanks!
15:30:34 <cdent> here's the whole set: https://review.openstack.org/#/q/topic:bp/event-alarm-evaluator,n,z
15:30:41 <cdent> all merged up now
15:30:57 <rhochmuth> great, will look at that too
15:31:10 <rhochmuth> i think my clone of the repo predates this
15:31:12 <jimbaker> rhochmuth, so any streaming design should be supportable on map/reduce; support multiple tenants; and be able to globally consolidate information
15:31:33 <rhochmuth> jimbaker, yes that is what i think would work too
15:31:34 <jimbaker> (ie, what we see in storm and the current monasca work)
15:31:52 <rhochmuth> yeah, but i'm not pushing storm
15:32:01 <rhochmuth> i would like to see a new python alarm engine
15:32:10 <jimbaker> sure, only as supporting infrastructure. although on spark streaming ideally
15:32:33 <rhochmuth> i think spark is a good basis
15:32:39 <deklan> jimbaker: why spark ideally?
15:33:14 <rhochmuth> i was goiing to prototype a very crude alarm engine in Spark, just to see what it looked like
15:33:18 <jimbaker> deklan, better integration with kafka, cassandra; exactly once processing for reliable dashboards
15:33:26 <rhochmuth> i could possible even use the Aodh in that, ideally
15:33:53 <rhochmuth> deklan says i'm all talk
15:33:55 <jimbaker> rhochmuth, there might be some core that we could extract
15:34:03 <jimbaker> ;)
15:34:07 <rhochmuth> yes, that is what i'm hoping for
15:35:24 <rhochmuth> ok, action items then is to follow up with gordon, Ildikó, nec and others
15:35:27 <jimbaker> rhochmuth, do you want to spend a day working on this together? i have some availability next week (tues or wed). of after i get back from vacation
15:35:40 <rhochmuth> yes, i could do that,
15:35:43 <rhochmuth> how about the following week
15:36:06 <rhochmuth> cdent are you interested in working on aodh too?
15:36:29 <jimbaker> sounds good. next wed is probably best, but tues is also workable
15:36:59 <rhochmuth> i can't to wed, let's talk off-line, and see what works bext then
15:37:01 <cdent> rhochmuth: I already do, at least from a review standpoint. I haven't done much in the way of features, but I tend to manage all its devstack-related stuff.
15:37:14 <rhochmuth> cdent ok
15:37:27 <jimbaker> rhochmuth, ack
15:37:59 <rhochmuth> cdent i'll send an email and we can coordinate with ceilometer folks too
15:38:15 * cdent will look out for it
15:38:18 <rhochmuth> and will review the nec stuff off-line
15:38:21 <rhochmuth> cdent ok
15:38:44 <jimbaker> rhochmuth, do you have blueprints or other info on NEC?
15:38:59 <rhochmuth> jimbaker, sorry i don't
15:39:16 <cdent> jimbaker: this of any use: http://specs.openstack.org/openstack/ceilometer-specs/specs/liberty/event-alarm-evaluator.html
15:39:19 <cdent> ?
15:39:19 <rhochmuth> but, the reviews posted are for NEC, i beleive?
15:39:41 <jimbaker> rhochmuth, that looks useful. i will have to read in depth!
15:40:11 <jimbaker> i rather not have to speculate on what they might do. easier to read the ideas :)
15:40:20 <rhochmuth> ok, any other pressing agenda items, before jim talks about bleflood?
15:40:33 <rhochmuth> bugs, code commits, reviews???
15:41:04 <rhochmuth> jimbaker i believe the floor is yours, unless someone has got anything else
15:41:09 <deklan> rhochmuth: please +2 https://review.openstack.org/#/c/218495/
15:41:15 <jimbaker> rhochmuth, thanks
15:41:53 <jimbaker> blueflood is a timeseries database for metrics that wraps cassandra. it is opensourced under an apache license
15:42:06 <jimbaker> we are currently ingesting 100K metrics/second
15:42:38 <jimbaker> we support kafka, scribe, statsd, and REST for metrics
15:42:44 <rhochmuth> that's pretty good
15:42:56 <bklei_> how's query performance?  any numbers there?
15:43:20 <jimbaker> then also support auto aggregation at 5 min, 60 min, 240 min, and 1 day intervals
15:43:51 <deklan> jimbaker: are you suggesting that we incorporate blueflood into monasca?
15:43:55 <jimbaker> bklei_, i don't have that at my  fingertips... would like to have alex scammon here to discuss, and he will be here next week
15:44:11 <jimbaker> but it is quite reasonable
15:44:18 <bklei_> sweet, very interested in that
15:44:26 <rhochmuth> yes, he was interested in doing this
15:44:28 <jimbaker> we also have grafana integration
15:44:36 <jimbaker> although not yet 2.x
15:45:04 <bklei_> what's the authentication story for grafana/blueflood integration?
15:45:07 <bklei_> keystone?
15:45:11 <jimbaker> so it's a pretty straightforward wrapping of cassandra. but deklan, i think we have done some nice value add
15:45:34 <deklan> jimbaker: what is the value add?
15:46:00 <jimbaker> we have the ingesters, the rollups, the use of elasticsearch to support globbing metrics
15:46:25 <jimbaker> so it's not just raw cassandra. but we don't slow cassandra down :)
15:46:49 <rhochmuth> does blueflood support something similar to dimensions in monasca?
15:46:56 <deklan> is anyone else in openstack using blueflood currently?
15:47:42 <jimbaker> bklei_, re the specific auth - i don't know details. i believe it starts with repose. could readily find details
15:48:19 <bklei_> k, would hope we integrate with keystone to support project level access
15:48:23 <jimbaker> deklan, not to my knowledge. but the 100K metrics/second that i quote is what we are currently supporting for our customers
15:49:02 <jimbaker> bklei_, it's multitenant. just don't know specifics on how we integrate with keystone
15:49:14 <bklei_> understood, that sounds good
15:49:53 <fabiog> jimbaker: bklei_ the big tend requirement is that is at least integrated with Keystone :-)
15:50:04 <deklan> jimbaker: any movement to make it an official openstack project?
15:50:13 <bklei_> +1 fabiog
15:50:44 <jimbaker> so it's not perfect. we plan to add support for metrics 2.0. but cassandra helps with collection columns. i have also started work on spark/spark streaming integration
15:51:02 <jimbaker> deklan, i don't know. the code base is completely in java
15:51:11 <deklan> lol
15:51:13 <jimbaker> this is for people who care about scale
15:51:30 <rhochmuth> jimbaker is the proposal to add blueflook as another bvackend option to monasca
15:51:37 <jimbaker> rhochmuth, exactly
15:52:10 <rhochmuth> jimbaker so, i don't see any problems with that
15:52:16 <mroderus> jimbaker: do you know for how long the blueflood project has existed so far?
15:52:18 <jimbaker> you should be able to use blueflood to get the most scalable, globally distributed support for monasca
15:52:33 <bklei_> would like to experiment with that for sure
15:52:58 <jimbaker> mroderus, it was a key part of rackspace's acquisition of cloudkick iirc
15:53:25 <jimbaker> then we opensourced it in 2013
15:53:29 <rhochmuth> jimbaker so if rx is interested in this then i think we would support
15:53:42 <rhochmuth> definitely woudl like to discuss further
15:53:58 <rhochmuth> we would like to add to the Python API implementation
15:54:09 <rhochmuth> if proceed
15:54:57 <rhochmuth> jimbaker next steps?
15:55:15 <jimbaker> sure, not certain what a python api here means. i guess one could have a python equivalent to the blueflood code, but that sort of defeats its goal of all-out performance
15:55:40 <jimbaker> again that's why i describe it as being a possible backend
15:55:40 <fabiog> jimbaker: can you post a link with documentation/design info on blue flood, please?
15:55:48 <jimbaker> http://blueflood.io/
15:55:55 <fabiog> jimbaker: thanks
15:55:57 <mroderus> would adding blueflood replace our existing Cassandra plans?
15:56:30 <jimbaker> mroderus, yes, that's my suggestion
15:56:45 <mroderus> ok
15:57:39 <rhochmuth> i think we'll need some serious discussion on this topic
15:57:46 <rhochmuth> ok, meeting is coming to a close
15:58:02 <jimbaker> rhochmuth, sounds good. i guess that was not quite so brief an intro
15:58:11 <jimbaker> but alex scammon can say more, hopefully next week?
15:58:15 <rhochmuth> yes, that was short
15:58:24 <rhochmuth> ok, let's discuss more next week
15:58:32 <jimbaker> rhochmuth, thanks
15:58:34 <rhochmuth> any other quick comments in closing?
15:59:03 <rhochmuth> ok, there are some action items, and follow-ups going on
15:59:20 <rhochmuth> not sure how to record them all, but please follow-up
15:59:28 <rhochmuth> i'm not exactly tracking
15:59:37 <jimbaker> it's in the log
15:59:38 <rhochmuth> if i do that i'll become a manager
15:59:43 <jimbaker> don't :)
15:59:48 <rhochmuth> lol
16:00:04 <deklan> a brain is a terrible thing to waste
16:00:16 <rhochmuth> allright everyone, meeting is closing
16:00:22 <deklan> ciao
16:00:23 <witek> rhochmuth: thanx
16:00:25 <jimbaker> later
16:00:29 <bklei_> thx rhochmuth
16:00:30 <rhochmuth> #endmeeting