15:00:04 #startmeeting monasca 15:00:06 Meeting started Wed Dec 14 15:00:04 2016 UTC and is due to finish in 60 minutes. The chair is rhochmuth. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:07 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:08 o/ 15:00:10 The meeting name has been set to 'monasca' 15:00:14 https://etherpad.openstack.org/p/monasca-team-meeting-agenda 15:00:16 o/ 15:00:20 o/ 15:00:22 Agenda for Wednesday December 14 2016 (15:00 UTC) 15:00:22 1. Grafana - meeting with Raintank 15:00:22 2. Reviews: 15:00:22 1. https://review.openstack.org/#/c/408813/ 15:00:22 3. Log API future for reads 15:00:22 4. Can we uncap python-kafka ? http://lists.openstack.org/pipermail/openstack-dev/2016-December/108436.html 15:00:23 5. New version numbers for Java components https://review.openstack.org/#/q/projects:openstack/monasca+topic:version 15:00:23 6. Ocata 2 Milestone: 15 Dec 15:00:24 7. Adding Promethues features to Monasca 15:00:24 1. PromQL 15:00:25 2. Alarms and notifications 15:00:25 1. https://prometheus.io/docs/alerting/alertmanager/ 15:00:26 o/ 15:00:26 2. Grouping Alerts 15:00:26 3. Alarm inhibition 15:00:28 o/ 15:00:31 o/ 15:00:33 o/ 15:00:35 o/ 15:00:39 hi everyone 15:00:57 looks like a good agenda today 15:00:59 hi all 15:01:07 and we have a number of people here which is great 15:01:17 i'll probably be out the next two weeks though 15:01:42 so, for next week, unless i'm at home at 8:00 AM doing nothing, i probably won't attend 15:01:55 but, you can never rule me out 15:02:05 so, someone else should start the meeting next week 15:02:25 so, let's go through the agenda 15:02:33 #topic Grafana - meeting with Raintank 15:02:41 That's me 15:02:42 rbak: is that u 15:03:06 Anyone who asked to be on last weeks email to Raintank may have already seen this 15:03:19 But we've setup a meeting with them for anyone who wants to attend. 15:03:35 It's at 1630 UTC, so half an hour after this meeting ends 15:03:56 where? 15:04:01 I don't have the link for the the meeting yet but it should be in the email and I'll post it here once I have it. 15:04:24 They said they would send one out to us. 15:04:30 https://plus.google.com/hangouts/_/raintank.io/openstack?hceid=ZHV0dEByYWludGFuay5pbw.pt93p3n8dpqq55i410e81c3aks 15:04:31 Here's the link for the meeting: https://plus.google.com/hangouts/_/raintank.io/openstack?hceid=ZHV0dEByYWludGFuay5pbw.pt93p3n8dpqq55i410e81c3aks 15:04:37 Great, thanks 15:05:35 That's all I had. Anyone is welcome to attend who wants to talk with them about Grafana. 15:05:49 One thing to mention up front: 15:06:15 dhague: are we going to try and keep the kubernetes meeting after this meeting 15:06:32 They seem to be of the mindset that openstack support will be a support burden for which they are looking for subscriptions from users (i.e. us), instead of seeing it as a contribution from us which will help their product 15:06:57 So we may have some perusading to do, or at least compromising. 15:07:05 Good point. 15:07:41 I'll add that they've started talking to Charter again, so I wouldn't get hopes up but we might be able to offer some leverage there. 15:07:58 yes, they are a business 15:08:21 o\ 15:08:28 so, let's see what they say later today 15:09:06 dhague: are we still trying to meet after this meeting? 15:09:19 rhochmuth: about the k8s meeting, jobrs rescheduled it for next week to avoid conflicts, but we could also move it to the new year if that's more convenient 15:09:32 If there is still resistance, would it be worth suggesting a plugin-style approach, and maintain a plugin outside grafana? 15:09:48 thanks dhague 15:09:57 stevejims: I think they're thinking pluggable auth would be a good idea. 15:10:09 stevejims: that would be great, except that Go doesn't really do plugins nicely 15:10:16 Ok, that's good 15:10:19 not sure i wanted hoppal to have to wait that long, as he is in the midst of developign his plugin 15:10:34 Right, so I don't know how they're planning on doing the pluggable auth. 15:10:41 We'll see what they say in the meeting. 15:11:01 I can do whenever works for everyone else 15:11:10 but as far as i'm concerned to the new year is fine with me 15:11:11 Sounds good 15:11:18 split brain 15:11:46 rhochmuth hoppalm: ok, I'll get jobrs to push it to the new year 15:12:12 ok, sorry for the change in plans, was hoping to resolve this 15:12:17 summary: prometheus mtg in new year, grafana meeting in 1:15 from now 15:12:39 hoppal's promethues and kubernetes plugins are up for review if you have some comments 15:12:44 correct 15:12:48 dhague 15:13:12 #topic review https://review.openstack.org/#/c/408813/ 15:13:24 Also mine. 15:13:35 Just wanted to get some eyes on. 15:13:46 It's been up for a week and only one comment. 15:14:11 looks like craig has +1'd 15:14:21 and bklei 15:14:29 so, i'm assuming ready to go 15:14:30 It's a pretty simple patch 15:14:52 if anyone else want to take a quick look and merge that would be great 15:15:02 or add comments as appropriate 15:15:26 #topic Log API future for reads 15:15:55 This was my addition 15:16:10 I was hoping to raise the topic of the future of the log API 15:16:38 hi stevejms, welcome to the monasca meeting 15:16:49 does anyone want to take a crack at this one? 15:16:51 Hi, thanks 15:16:52 we thought of doing this, but for now decided to implement multi-tenancy in Kibana plugin 15:16:53 if not, i can 15:17:31 We have looked at the Kibana plugin but it looks like the MT was reverted? 15:17:32 so, obviousely we would like to see this added 15:17:36 it isn't trivial 15:17:59 do to the prioritues and resources we don't have any immediate plans to add support for reading logs from the log api 15:18:15 we are in the process of adding this to devstack plugin 15:18:47 https://github.com/FujitsuEnablingSoftwareTechnologyGmbH/fts-keystone/tree/multitenancy 15:19:02 the plan is: 15:19:21 * update tempest tests for logs 15:19:34 * create new repo for plugin 15:19:42 * include in devstack-plugin 15:20:25 Sounds promising - Would be there be interest in proposals for an API, if we (I) could put some time into it? 15:21:00 The use case would be allowing logs to be displayed in Grafana, so quite focused 15:21:08 i think the answer is yes, we are always interested in developers becoming involved with the project 15:21:31 but, we also need to coordinate 15:21:35 witek? 15:22:00 sure 15:22:14 Also, I want to mention, if we ever get to adding reads to the log api, we would want to look at the Searchlight for code reuse/inspiration… 15:22:15 we can work together on API proposal 15:23:09 Searchlight has an api for quierying ES, that handle multi-tenancy, filtering of in/out bound queries, and making sure that queries are secure 15:23:29 Sounds good - happy to coordinate on a proposal 15:23:42 stevejims: so my suggestion would be to submit a blueprint first 15:23:50 along with some API proposals 15:24:06 i'm wondering if we want to try and use storyboard for something like this, rather than launchpad 15:24:37 good idea 15:24:42 then we can review either in weekly meetings or in a separate time/place 15:24:49 witek: storyboard? 15:24:51 Sure 15:24:53 yes 15:25:13 stevejims: are you familiar with storyboard? 15:25:29 Afraid not but can get familiar 15:25:43 https://storyboard.openstack.org/#!/page/about 15:26:15 we need to add a monasca project to it i think 15:26:18 hopefully that isn't hard 15:26:46 so let's close on this for now 15:26:54 i'll assume we'll give storyboard a try 15:27:39 stevejims, looks like you have approval to get started on a proposal and review with the team 15:27:40 OK grand - thanks all 15:27:56 where do you work, btw? 15:28:36 #topic Can we uncap python-kafka 15:28:51 http://lists.openstack.org/pipermail/openstack-dev/2016-December/108436.html 15:28:51 I have put that one 15:29:00 Company is called StackHPC, we primarily do consulting work for Cambridge University on OpenStack 15:29:17 cool, thanks 15:29:57 there is some pressure from requirements team and oslo.messaging to uncap python-kafka 15:30:25 my understanding after talking to joe about this is that it isn't a great situation 15:30:41 they want to bump the kafka library to the latest one 15:30:50 however, it only supports async 15:31:08 there is a SimpleConsumer, that is depreated in the library, that they want us to move to 15:31:24 but, there are a number of issues outstanding in SimpleConsumer too 15:31:41 i guess there is a SimpleProducer too 15:31:55 i would like to free the openstack oslo folks up too 15:32:17 but, this change puts us in a potentially bad situation 15:32:45 as an alternative, we were thinking about moving to the confluent library 15:33:05 then we could get out of everyone's way 15:33:34 thoughts? 15:33:45 confluent is not in global-requirements though, right? 15:33:57 right 15:34:08 we would have to get that added first 15:34:17 then we could move 15:34:32 do they support sync? 15:34:53 i don't think so, but i'm not positive 15:35:16 so, it isn't exactly a trivial change 15:35:57 joe is out for the rest of the year i believe 15:36:06 so, i can't check with him 15:37:16 is there anything else to discuss on this right now 15:37:31 ok, thanks for explaining, it seems complicated at that point 15:37:38 yes, it is 15:38:13 so, let's get back to this issue in the new year 15:38:21 #topic New version numbers for Java components 15:38:31 https://review.openstack.org/#/q/projects:openstack/monasca+topic:version 15:38:44 could we bump the version numbers of Java components? 15:39:03 sure 15:39:15 ok by us at charter 15:39:34 the tests should be passing if we merge in a right sequence 15:39:53 so, if you want to shepherd it through that would be great 15:40:21 monasca-common needs to go first i guess 15:40:27 yes 15:41:09 ok, i guess feel free to +2 15:41:12 then thresh and persister 15:41:26 right 15:42:28 #topic Ocata 2 Milestone: 15 Dec 15:42:39 just a reminder 15:42:41 omg 15:42:44 :) 15:42:46 already 15:43:03 I'll create tags 15:43:16 sounds great 15:44:12 #topic Adding Promethues features to Monasca 15:44:21 this is my topic 15:44:41 just wondering if anyone else had any thoughts or ideas on this 15:45:04 i have a little prometheus envy lately 15:45:47 but, at the same time, i think we have some compelling features, functionality and architecture with monasca 15:46:21 the prometheus community is starting to look at areas like scale, clustering, monitoring-as-a-service 15:46:22 it seems like everybody is getting in on the alerting act: prometheus, ceilometer, grafana, ... 15:46:48 yes, there is a huge amount of overlap 15:48:00 anyway, i'm looking at this area a bit, and trying to understand what could be or should be done 15:48:07 just wanted to mention that 15:48:40 leverage ro replacement? 15:48:43 or 15:49:00 i think understand first 15:49:15 we have a working student who is doing his thesis on monitoring our "converged cloud" openstack-on-kubernetes setup. His report should be available early in the new year, and hopefully we can share it here. It covers prometheus and monasca, among other things 15:49:30 my first thought is to try and add features to monasca to remain parity 15:49:37 thanks dhague 15:49:54 i also did a comparison, but unfortunatly on the internal wiki 15:49:58 a diff of the features would be useful. 15:50:10 so, i'll try and get that published on the monasca public wiki 15:50:44 When I have talked to people about Prometheus the thing they like the most is the simple nature of it, but that is also it's biggest issue (scalability) 15:50:45 thanks rhochmuth 15:51:06 i was going to mention that there are two companies, weaveworks and digitalocean that have leverage promethues 15:51:12 https://github.com/digitalocean/vulcan 15:51:34 https://github.com/weaveworks/cortex 15:52:37 ok, enough on that topic 15:52:48 #topic open floor 15:54:52 i guess no other topics 15:54:57 so, i'll end the meeting 15:55:03 thanks everyone 15:55:10 thanks 15:55:12 see some of you in a few minutes 15:55:25 thanks 15:55:28 #endmeeting