15:00:30 <rhochmuth> #startmeeting monasca
15:00:34 <openstack> Meeting started Wed Apr  6 15:00:30 2016 UTC and is due to finish in 60 minutes.  The chair is rhochmuth. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:35 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:37 <openstack> The meeting name has been set to 'monasca'
15:00:49 <rhochmuth> o/
15:00:55 <bklei> o/
15:00:55 <tomasztrebski> o/
15:00:59 <Kamil_> o/
15:00:59 <hosanai> o/
15:01:05 <shinya_kwbt> o/
15:01:08 <witek> hello
15:01:16 <rbak> o/
15:01:16 <rhochmuth> Link to agenda is at, https://etherpad.openstack.org/p/monasca-team-meeting-agenda
15:01:27 <rhochmuth> Agenda for Wednesday April 6, 2016 (15:00 UTC)
15:01:28 <rhochmuth> 1. Replace agent thread pool https://review.openstack.org/#/c/301355/
15:01:28 <rhochmuth> 2. monasca-transform
15:01:28 <rhochmuth> 3. monasca-analytics
15:01:28 <rhochmuth> 4. Summit scheduling
15:01:35 <rhochmuth> Hi everyone
15:01:43 <rhochmuth> It has been a crazy week for mee
15:01:46 <tgraichen> hi
15:02:18 <rhochmuth> There is a small agenda today, so it looks like we'll have time for other topics
15:02:31 <tomasztrebski> even crazier for me....
15:02:50 <rhochmuth> shinya, thank you for fixing the problem with the nodejs install in Grafana last night
15:03:01 <shinya_kwbt> Not at all
15:03:12 <rhochmuth> i've merged it and it is in flight
15:03:29 <slogan> I ended up simply doing a sudo apt-get install nodejs and it all worked, nothing complicated. What was the fix?
15:03:31 <shinya_kwbt> Thanks
15:03:41 <rhochmuth> i had done something similar myself yesterday, but hadn't posted it yet
15:04:09 <rhochmuth> The reivew hasnt' merged yet, but it is at, https://review.openstack.org/#/c/302077/
15:04:17 <slogan> thx
15:04:50 <rhochmuth> On a related topic, I tried the Monasca plugin for the Logging API/Service yesterday
15:05:12 <rhochmuth> The review is at, https://review.openstack.org/#/c/300396/
15:05:19 <rhochmuth> I haven't really looked at the code
15:05:34 <rhochmuth> just tried using it along with MOnasca
15:05:43 <rhochmuth> It appears to work
15:05:49 <rhochmuth> At least Kibana comes up
15:05:51 <witek> nice to hear! :)
15:05:54 <tomasztrebski> I saw you posted a comment about default index, did you manage to resolve that ?
15:06:00 <rhochmuth> I left a comment
15:06:04 <rhochmuth> no, i haven't resolved that
15:06:11 <rhochmuth> if you know the magic, let me know
15:06:15 <rhochmuth> please
15:06:32 <rhochmuth> so, i'm not sure it is working completely, but at least it installs and runs
15:06:40 <tomasztrebski> I'll respond to that, it just Kibana does not know what would like to display, so it needs an index name or a pattern to read from
15:06:53 <witek> the index pattern should be tennantid-YYYY-MM-DD
15:06:58 <rhochmuth> ok, thanks
15:07:09 <rhochmuth> i'll need to get tsv to start looking into it
15:07:36 <rhochmuth> it is really cool to see all this stuff, Monasca, Horixon, Grafana and ELK all running in the same DevStack VM
15:08:00 <rhochmuth> That is going to make the Bootcamp session on Monasca much more enjoyable
15:08:13 <rhochmuth> So, thanks everyone for helping to get that working
15:08:44 <rhochmuth> I'm assuming it will be used in other sessions too
15:09:07 <rhochmuth> Now, i need to get the Ceilosca stuff working
15:09:41 <rhochmuth> #topic Replace agent thread pool https://review.openstack.org/#/c/301355/
15:09:49 <rbak> That's me
15:09:56 <rhochmuth> yes
15:10:06 <rhochmuth> that u
15:10:09 <rbak> This patch has been up for a few days and I just wanted to get some eyes on it.
15:10:30 <rhochmuth> i know some folks on the team here at hpe have started to look at it
15:10:42 <rhochmuth> but, i see that they haven't left comments
15:10:51 <rhochmuth> we are pretty busy with a release right now
15:11:00 <rhochmuth> but, i'll ping them
15:11:02 <rbak> As long as someone's looking at it.
15:11:17 <rbak> The agent locking up has been a problem for us for a while
15:11:32 <rhochmuth> we were wondering if we wanted to pick it up for the release
15:11:50 <rhochmuth> how urgent is tis to resolve on your end
15:12:06 <rhochmuth> how soon are you looking to get merged
15:12:33 <rbak> I'd like to get it as soon as possible.  We're opening our service up to customers so they're going to see the holes in our data where agents locked up.
15:12:58 <rhochmuth> how long have you been testing this change
15:13:04 <rbak> But I know it's a big change, so I don't want to rush it if someone isn't comfortable
15:13:18 <slogan> I started looking at it a day or two ago, but got distracted - but looks good +1
15:13:20 <rhochmuth> well, i think that was the concern for us
15:13:35 <rbak> I've had it running for a couple days at least with no problems.
15:13:45 <rhochmuth> cool, that is good info
15:13:53 <rhochmuth> have you seen any memory growth
15:14:03 <rbak> And it does recover when checks get stuck.
15:14:04 <rhochmuth> resource leaks?
15:14:15 <rbak> Nothing I've noticed so far.
15:14:30 <rhochmuth> ok, i'll let joe and other know
15:14:42 <rhochmuth> i've added joe to the review and will add some others
15:14:48 <rbak> I haven't looked recently though, I'll go back and see how the resources are looking.
15:16:05 <rbak> I don't have anything more on that subject.  I'll post an update on the patch.
15:16:47 <rhochmuth> #topic monasca-transform
15:17:20 <rhochmuth> So, I've posted a review at, https://review.openstack.org/#/c/301313/, for monasca-transform
15:17:28 <rhochmuth> The monasca-transform project is a new component in Monasca that
15:17:28 <rhochmuth> will aggregate and transform metrics. For example, it will consume
15:17:28 <rhochmuth> metrics for Swift from the Monasca metrics topic and count the total
15:17:28 <rhochmuth> number of times a container or object is accessed and group them by
15:17:28 <rhochmuth> project. Then it will publish the aggregated metric to the Kafka
15:17:29 <rhochmuth> metrics topic where the rest of the existing Monasca components
15:17:29 <rhochmuth> will be used to store, alarm and notify.
15:17:55 <rhochmuth> This component has been in the architecture diagram for a long time, and we've talked about it in the past
15:18:08 <rhochmuth> But, it looks like we're going to get some resources fomr HPE to work on this
15:18:32 <rhochmuth> I would like to get this reviewed next week
15:18:38 <rhochmuth> If that is OK with everyone
15:19:00 <rhochmuth> Basically, go over uses cases, architecture, design, …
15:19:44 <rhochmuth> The goal is to complete this component by the Newton release or sooner
15:19:49 <tomasztrebski> is it this one -> https://github.com/hpcloud-mon/monasca-events-transform ?
15:20:02 <rhochmuth> No, that was targeted at events
15:20:05 <rhochmuth> This is for metrics
15:20:14 <rhochmuth> Totally new repo
15:20:18 <rhochmuth> will be created
15:20:21 <slogan> the use cases would be interesting to read
15:20:30 <slogan> where can we find more info?
15:20:53 <rhochmuth> It is being created and will publish and review next week
15:21:06 <slogan> k
15:21:15 <rhochmuth> They've created proto code
15:21:53 <rhochmuth> So, are folks OK covering this next week, or shoudl we arrange another time-slot
15:22:01 <rhochmuth> it might take a while to get through the entire topic
15:22:57 <witek> some document would be nice
15:23:11 <rhochmuth> ok, it is on the way
15:23:21 <rhochmuth> hopefully, others will want to work on this too
15:23:29 <tomasztrebski> ]:->
15:23:39 <rhochmuth> what does that mean?
15:24:10 <tomasztrebski> nothing bad, I would want to see how this one work and fits in everything else :D
15:24:29 <rhochmuth> yes, that is the plan
15:25:07 <rhochmuth> #topic monasca-analytics
15:25:44 <rhochmuth> So, I'll also be submitting a request for a new repo for the alarm clustering and anomaly detection work that we've been starting to discuss again
15:26:09 <hosanai> nice :-)
15:26:13 <rhochmuth> As soon as I get monasca-transform done, i'll submite monasca-analytics
15:26:19 <rhochmuth> which shoudl go smooth
15:26:29 <rhochmuth> hosanai, sorry about the delay
15:26:42 <hosanai> rhochmuth: np!
15:26:43 <rhochmuth> but, i'm getting to it
15:27:08 <rhochmuth> at some point, we can also do a similar review, but much less is known about this area right now
15:27:23 <rhochmuth> materials are being created, but it is much more complicated
15:28:02 <rhochmuth> curretly, hosanai, a small team in hpe, and me are interested in this area
15:28:17 <rhochmuth> so, look for that
15:28:28 <slogan> what is alarm clustering?
15:28:28 <rhochmuth> i just didnt' want to surprise anyone
15:28:44 <rhochmuth> hosanai, do you want to explain?
15:28:51 <slogan> I we talking k-means type of thing?
15:28:59 <tomasztrebski> maybe once you have all these materials available (use cases, documentation of some sort), you could post a mail via mailing list to notify everyone, just when it is ready
15:29:16 <rhochmuth> sure tomasz, will do
15:29:21 <tomasztrebski> thx
15:29:33 <slogan> that would be good
15:29:53 <slogan> I'm so focused on the lowest level, I need to surface and see the bigger picture
15:29:59 <hosanai> ok, my typing speed is not enough
15:29:59 <rhochmuth> hosanai, can you explain alarm clustering, or do you want me to describe?
15:30:17 <hosanai> rhochmuth: could you explain it?
15:30:23 <rhochmuth> sure
15:30:55 <rhochmuth> currently, many alarms can occur simultaneousely
15:31:06 <rhochmuth> For example, if a system goes down
15:31:19 <rhochmuth> all the alarms on that system, will fire
15:31:56 <rhochmuth> The idea is to cluster alarms temporally and based on other features of the metrics/alarms
15:32:10 <rhochmuth> to reduce the number of alarms and reduce alarm fatigue
15:32:33 <rhochmuth> So, instead of 100s of alarms, you would get a few.
15:32:36 <slogan> I see, so, selected attributes and ranges define membership in a cluster
15:32:52 <rhochmuth> Correct
15:32:56 <slogan> you get told about the cluster, and can peel the onion for details later
15:33:02 <rhochmuth> K-means woudl be one way to do the clustering
15:33:14 <rhochmuth> but, i don't think that is what is planned right now
15:33:23 <slogan> SOM :-)
15:33:29 <rhochmuth> THere are two companies that do this that i'm aware of
15:33:33 <rhochmuth> moogsoftware
15:33:36 <rhochmuth> Big Pandas
15:34:13 <rhochmuth> #topic Summit scheduling
15:34:41 <rhochmuth> Currently, we have five sesions at the design summit for Monasca related discussions
15:34:58 <rhochmuth> All are Wednesday from 9:00 - around 1:00
15:35:03 <rhochmuth> There is a stop for lunch
15:35:22 <slogan> 40 minute or so sessions I guess
15:35:29 <rhochmuth> Correct
15:35:47 <rhochmuth> The first session will be general overview, planning and roadmap
15:36:30 <rhochmuth> The second session will be more deetailed discussions on specific features, siuch as multiplke metrics, non-periodic metrics, periodic notifications, retention periods, …
15:36:42 <rhochmuth> Third session will be logging API/service
15:36:57 <rhochmuth> Fourth session will be Networking/Neutron/Broadview
15:37:13 <rhochmuth> Fifth session is still available
15:37:26 <rhochmuth> Sorry, should have said that is my proposal, at this point
15:37:37 <rhochmuth> Nothing has been defined and set in stone
15:37:54 <rhochmuth> So, I'm wondering if that sounds reasonable
15:38:10 <rhochmuth> I know a lot of folks won't be there
15:38:26 <witek> sounds good for me
15:38:30 <slogan> sounds reasonable, I'm more likely to get travel permission if it becomes set in stone.
15:38:38 <slogan> *sigh*
15:38:56 <rhochmuth> I'll check with Armando on the Networking/Neutron discussion
15:38:57 <slogan> (that sigh is with my side, not you)
15:39:17 <rhochmuth> But, i think it is all set
15:39:25 <rhochmuth> i'll confirm
15:39:47 <slogan> I'll go ahead and tell my manager I have to be there, you can back me up if needed
15:39:59 <slogan> we need this discussion to happen
15:40:20 <rhochmuth> i think you should attend all the Neutron sessions, which there are a lot of, and thety don't overlap with Monasca
15:40:39 <slogan> nod
15:40:48 <rhochmuth> Interfacing with that team would be great
15:41:13 <slogan> totally agree
15:42:05 <rhochmuth> so, unless i hear otherwise, that is the working plan at the moment
15:42:33 <rhochmuth> hosanai, i was wondering about using the fifth slot for anomaly/analytics related discussions
15:42:43 <rhochmuth> i'm just not sure who else is interested in that topic
15:43:01 <rhochmuth> i am up for discussions while you are there
15:43:17 <hosanai> rhochmuth: that is the point. i'm availeble in there
15:44:16 <hosanai> if we have 5th slot and get more attention for it. we can have the slot. maybe...
15:44:57 <rhochmuth> i agree,
15:45:29 <rhochmuth> but usually, you need the interest first, before scheduling the session
15:45:55 <rhochmuth> i could leave the session open, with analytics as one of the topics to discuss in more detail
15:46:00 <rhochmuth> dependent on interest
15:46:20 <hosanai> sounds nice!
15:46:33 <rhochmuth> so, that sounds like a reasonable plan
15:46:52 <rhochmuth> i'll leave it an open session with some suggested potential topics dependent on interest
15:47:00 <rhochmuth> and we can refine while we are there
15:47:41 <rhochmuth> sound good?
15:47:51 <hosanai> +1
15:47:56 <rhochmuth> thx
15:48:06 <rhochmuth> So, we've got around 10 minutes left
15:48:13 <rhochmuth> i'll just open it up for discussion
15:48:26 <rhochmuth> #topic general discussion
15:49:00 <tomasztrebski> I wanted to clarify this change => https://review.openstack.org/#/c/301443/
15:49:02 <witek> it would be great if Craig could have a look at our monasca-thresh chage
15:49:10 <tomasztrebski> and that too
15:49:14 <tomasztrebski> ;D
15:50:02 <rhochmuth> craig has been extremely busy on a release, so his reviews have been a bit delayed
15:50:07 <rhochmuth> as have mine
15:50:23 <rhochmuth> so, i'm hoping he'll start to free up here today/tomorrow
15:50:41 <rhochmuth> i'll ping him too
15:50:52 <witek> thank you
15:51:00 <tomasztrebski> +1
15:52:11 <rhochmuth> so, are there other topics
15:52:31 <tomasztrebski> I've posted a link above to switch to different partitioner in Kafka
15:52:58 <tomasztrebski> since current one is deprecated due to certain bug (#374) in python-kafka
15:53:25 <rhochmuth> that looks interesting
15:53:49 <rhochmuth> joe, will want to take a look at that
15:53:57 <slogan> rhochmuth: I ust got my travel approved, definitely there
15:54:05 <rhochmuth> thanks slogan
15:54:13 <rhochmuth> maybe it was my comments in this irc session?
15:54:15 <rhochmuth> :-)
15:54:24 <slogan> I pasted them in the e-mail, yes
15:54:35 <slogan> :-)
15:54:56 <rhochmuth> Also, I wanted to point out this review again, https://review.openstack.org/#/c/289675/
15:55:15 <rhochmuth> Ryan Brandt has been working on that one, and I beleive he thinks it is ready to go
15:55:25 <rhochmuth> Support for both InfluxDB and Vertica are complete
15:55:40 <rhochmuth> Although, the review is in merge conflict right now
15:57:03 <rhochmuth> On another note, just want to mention that we are not proceeding with Cassandra support at this point
15:57:40 <rhochmuth> We will take a wait and see approach, and revisit based on customer interest
15:57:40 <witek> why?
15:57:51 <rhochmuth> it is a lot of work
15:58:03 <rhochmuth> influxdb satisfies the open-source requirement
15:58:31 <witek> thanks for info
15:58:50 <rhochmuth> if anyone else wnat to work on, we would be completely opne to the submission
15:59:08 <rhochmuth> so, we're not going to prevent development in this area
15:59:18 <rhochmuth> we're just not going to be able to staff it right now
15:59:35 <shinya_kwbt> I want to try
15:59:48 <rhochmuth> shinya: you want to try cassandr?
16:00:17 <rhochmuth> ok, i need to end the meeting?
16:00:19 <shinya_kwbt> Yes But I don't know cassandra so I must study hard
16:00:32 <rhochmuth> ok, sounds good
16:00:40 <rhochmuth> there were some reviews that were in progress
16:00:43 <rhochmuth> got to go now
16:00:48 <Kamil_> bye
16:00:53 <hosanai> thanks& bye
16:00:54 <rhochmuth> #endmeeting monasca