16:00:08 #startmeeting Performance Team 16:00:09 Meeting started Tue Jun 28 16:00:08 2016 UTC and is due to finish in 60 minutes. The chair is DinaBelova. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:11 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:00:13 The meeting name has been set to 'performance_team' 16:00:34 AugieMena o/ 16:00:39 o/ 16:00:43 hockeynut o/ 16:00:51 ok, so let's wait for a few moments 16:01:27 frankly speaking I do not have much to share today due to the final steps mirantis team is doing for q2 researches 16:01:32 but anyway :) 16:02:04 unfortunately same here - holiday week coming up here in the USofA 16:02:15 hockeynut yeah, I know this 16:02:48 ok, so let's get started 16:03:07 we did not have action items last time, so let's jump directly to the test plans 16:03:11 #topic Current progress on the planned tests 16:03:43 right now Mirantis is finishing publishing test plans and results to the performance-docs 16:03:52 #link https://review.openstack.org/#/q/project:openstack/performance-docs+status:open 16:04:18 AugieMena, hockeynut - folks, I'm really encouraging you folks to review the changes 16:04:33 I've add it to my list 16:04:38 this is something we're going to publish, so external feedback is really appeciated 16:04:41 hockeynut thank you sir 16:04:46 mon pleasure! 16:04:50 :) 16:04:56 took a quick look at one earlier this morning 16:05:15 in short there is one thing left we're doing right now - more realistic 1000 nodes emulation 16:05:47 we have 250 HW nodes with several VMs per node we're using as virtual nodes for 1000 nodes cluster 16:06:03 we have just installed the OpenStack cluster - benchmarking is coming 16:06:20 we're going to run usual dataplane and controlplane tests against it 16:06:26 to see what issues will be met 16:07:36 Very first assumptions on scale of 1000 nodes will be the following: it would 16:07:37 be good to have 2 dedicated servers per component. Here is a list of components 16:07:38 whose would require that: nova-conductor,nova-api, neutron-server, keystone. 16:07:39 RabbitMQ and MySQL servers worked in standalone mode so clustering overhead 16:07:40 will be added and they will consume much more resources than we already 16:07:41 metered. 16:08:03 this is something we met during very first emulation - https://review.openstack.org/#/c/324806/5/doc/source/test_results/1000_nodes/index.rst 16:08:14 let's see what we'll find out now 16:08:20 everything else is already on review 16:08:39 so that's all for current progress on the test plans 16:08:48 the test results are out there - are the tests themselves available? 16:09:08 hockeynut - do you mean for 1000 nodes? 16:09:11 sure 16:09:39 Alexandr Nevenchannyy proposed openstack/performance-docs: Add reliability test results https://review.openstack.org/333484 16:09:53 the idea is that we might be able to create a standard set of tests (think refstack) and these big ones could be a part of that 16:10:51 hockeynut yes, sure - so very first primary emulation is presented here https://review.openstack.org/#/c/324806/5/doc/source/test_plans/1000_nodes/plan.rst - dataplane and control plane are less specific we're running the same set of tests against every env - https://review.openstack.org/#/c/330629/ and http://docs.openstack.org/developer/performance-docs/test_plans/tenant_networking/plan.html 16:11:25 excellent - now I have more homework to do :-) 16:11:36 so during primary emulation we just ensured everything except computes themselves seem to be ok 16:11:53 and dataplane / control plane is something we want to be pretty stable for every env testing 16:12:03 so we came up with those two ^^ 16:12:12 but we can discuss what can be added / etc. 16:12:17 comments are very welcome :) 16:12:22 hockeynut indeed :) 16:12:29 have a nice homework :D 16:12:35 oh yes, I shall! 16:13:06 and in fact that's all about current test plans and the progress 16:13:14 plans for q3 are under creation 16:13:27 once I'll have a draft I'll share it here 16:13:36 so please be available this week :) 16:13:49 I think your feedback will be really improtant 16:13:54 AugieMena, hockeynut ^^ 16:14:06 got my reading glasses all polished up and ready to go 16:14:11 :D 16:14:16 ok, will do 16:14:42 ok, so as rohanion was ill during previous week I have nothing to share in the OSprofiler part 16:14:54 so I'll jump directly to the open discussions 16:14:55 #topic Open Discussion 16:15:03 in fact I have something to share here 16:15:14 I'm going to relocate to USA on July 16th 16:15:24 wow! 16:15:33 ooh - congrats! #dinaexit 16:15:36 and most probably I won't be able to hold the meeting July 12 16:15:38 :) 16:15:49 where? its a big country... 16:15:55 CA :) 16:16:03 one of the nicer parts. good choice 16:16:09 around Sunnyvale 16:16:15 very nice... 16:16:16 I am in Austin, TX - its like CA but cheaper and hotter 16:16:22 hockeynut I was there :) 16:16:25 on summit :) 16:16:45 really hot this spring there - I'm afraid of summer temperatures :D 16:16:54 not to jump the gun too much but any thought on Ocata items? 16:17:29 hockeynut I want to propose session around our researches 16:17:41 but did not pick up the exact item to talk about 16:17:43 yet 16:18:00 Alexandr Nevenchannyy proposed openstack/performance-docs: Add reliability test results https://review.openstack.org/333484 16:18:11 because overlook session is something I did last time, probably this summit it'll be a good idea to focus on something more specific 16:18:28 probably 1000 nodes benchmarks will be an interesting topic to cover 16:18:35 oh yes 16:18:49 so I'll propose a session soon 16:19:19 also I'll try to catch a separated time slot for performance team discussions 16:19:31 not a session, but just free communication format 16:19:46 but not sure yet if this will be possible - I'll keep you updated 16:20:14 hockeynut do you have any ideas what might be interesting? 16:21:11 AugieMena your ideas are very welcome too :) 16:21:17 :) 16:21:54 anyway you don't have to answer now, but if there will be any ideas - please ping me 16:22:04 combined sessions is something nice to have as well 16:22:24 so I really want to encourage people to propose their ideas and results 16:22:37 thinking about what we are doing in OSIC and there must be synergy here 16:23:18 the format in the session from Austin seemed fine, i.e., describing work items and results, open discussion on performance work by attendees there (who may not join these on a regular basis) 16:23:18 hockeynut indeed 16:23:36 AugieMena ack, so I'll add the overlook session as well 16:23:38 AugieMena thanks 16:23:57 ok :) 16:24:00 ok, anything else? 16:24:43 it looks like nope 16:24:45 not I 16:24:52 nor I 16:25:01 AugieMena hockeynut thanks for joining - please proceed with the reviews 16:25:01 :) 16:25:11 of course! 16:25:15 yep 16:25:18 do not afraid to set -1s :) 16:25:22 I love -1s :D 16:25:31 bye! 16:25:32 #endmeeting