15:30:01 #startmeeting Performance Team 15:30:02 Meeting started Tue Oct 10 15:30:01 2017 UTC and is due to finish in 60 minutes. The chair is DinaBelova. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:30:03 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:30:05 The meeting name has been set to 'performance_team' 15:30:15 ilyashakhat o/ 15:30:33 DinaBelova, hi/ 15:30:54 hm, it looks like we do not have any other folks sin irc channel so far :) 15:31:05 let's wait for a few moments 15:32:31 msimonin hey o/ 15:32:42 hey DinaBelova :) 15:32:50 you are here :p 15:32:57 :D you too :D 15:33:00 ll 15:33:03 lol* 15:33:28 okay, it looks like we have ilyashakhat and msimonin - sad that I do not see tovin07 so far 15:33:52 hm, well, it looks like we'll have kinda quick session now 15:34:11 #topic Action Items 15:34:33 do we have any? 15:34:42 last time we had only one action item on tovin07, so lemme keep it this time as well, as he does not seem to join today 15:35:01 #action tovin07 identify how to separate call/cast rpc calls within osprofiler tracing 15:35:17 so we can proceed for now 15:35:21 #topic Current status 15:35:35 msimonin I see you've uploaded new version of the test plan on review :) 15:35:49 msimonin anyway - what's current status on your side? :) 15:35:58 oh yes we made one more iteration with kgiusti 15:36:10 This test plan is ready for reviews 15:36:21 #link https://review.openstack.org/#/c/491818 15:36:27 #action DinaBelova ilyashakhat review https://review.openstack.org/#/c/491818/ 15:36:36 Now we'll start implementing it 15:36:53 with rmq and qpid-dispatch 15:38:18 msimonin as it looks like you're only person from inria today - any news on performance study of the databases ? including cockroachDB research? 15:38:58 yes rcherrueau won't join today 15:39:01 not too much 15:39:29 msimonin ack 15:39:37 I think we need to formalize a plan as well 15:39:48 ilyashakhat so let's switch to you, sir :) any news on current progress? 15:39:57 msimonin ack, let us review it 15:40:08 nothing new from my side, sorry :( 15:40:20 msimonin I believe we'll give some feedback this week 15:41:03 oh great :) 15:41:06 msimonin anything else to share, sir? 15:41:26 we had a pad about Q release plan 15:41:40 #link https://etherpad.openstack.org/p/performance-team-q-release-ideas 15:42:28 oh they are some +1 :) 15:42:53 From my side there are things related to fog/edge 15:43:01 line 11-13 15:43:06 yeah, from plans perspective it looks like we have a pack of things that you folks are planning to do (you're going to do it anyway, but i'll set my +1s there :D) 15:43:15 yes 15:43:22 I've put a note at the end 15:43:29 about "continuous benchmarking tool" 15:43:39 oh, I missed this particular item 15:43:49 I'm curious to know if it has already been discussed in the OpenStack community 15:44:02 hm, I did not see much about this 15:44:31 if we'll skip what rajulk2 is doing 15:44:49 they even have a name for a project: https://openstack-lcoo.atlassian.net/wiki/spaces/LCOO/pages/13393034/Eris+-+Extreme+Testing+Framework+for+OpenStack 15:44:56 I've copy/paste the comment from mriedem 15:45:08 BTW, DinaBelova do we have any relations with LCOO team? 15:45:28 ilyashakhat I believe not that much 15:46:21 as I understand they took some ideas from Timur's session in Barcelona and moved forward with it 15:46:36 msimonin I see. Well, from scale testing perspective my folks swithced to virtualized envs for the scale testing, this means that high scale + high number of VMs is not anymore possible on our virtualized labs 15:47:23 ilyashakhat it looks like this, I see stepler among the tools to be used 15:47:57 yep, and I even saw a report that looks similar to reliability that I made last year 15:48:02 DinaBelova: you mean you don't have access to bare metal ressources anymore ? 15:48:06 a moment, will find a link again 15:49:06 msimonin yes, for large scale. We do have specific HW labs for particular performance researches, but it's not about high-scale bm 15:49:37 it's kinda costy, and yet mostly deployment architecture is usually verified, so it's ok to run those tests on vms usually 15:49:56 Ok I see 15:51:29 btw, do we have any plan to evaluate the CellV2 ? 15:51:43 DinaBelova, will you be interested in measuring performance impact of profiling? 15:52:17 msimonin not from my side, my team will be kinda busy with other items to be covered 15:52:27 ilyashakhat I like this item, yes 15:52:34 ilyashakhat: me too :) 15:52:45 cool :) 15:53:06 ilyashakhat but I guess I can provide very little of myself to participate in this :D 15:53:07 I believe tovin was also interested 15:53:11 even in taing active part 15:53:15 Do you envision the way you can evaluate the impact ? 15:53:16 taking* 15:54:44 I would start with comparing duration of long operations like VM creation 15:55:02 using Rally ? 15:55:25 + resource consumption of openstack services and the osprofiler storage 15:55:36 Rally can be an option 15:56:22 profiling can be enabled directly in Rally scenario, so it's easy to use now 15:57:06 ilyashakhat i believe it will be the most convinient option now, yes 15:58:57 here's the HA testing report from LCOO: https://openstack-lcoo.atlassian.net/wiki/download/attachments/13393034/TestResultsAnalysis-2.docx?version=1&modificationDate=1500988830906&cacheVersion=1&api=v2 15:59:11 quite impressive 15:59:52 ilyashakhat it looks right like you did with reliability testing once :) 15:59:54 I mean charts 16:00:01 yep 16:00:12 looks really nice, I agree 16:00:15 thanks for findind it :) 16:00:22 finding* 16:00:32 I keep an eye on them :) 16:00:44 cool 16:01:06 I guess let's go back to https://etherpad.openstack.org/p/performance-team-q-release-ideas next time once tovin will join uss next time 16:01:25 ack 16:01:32 #topic Open Discussion 16:01:48 okay, any more interesting items to cover? 16:02:22 ilyashakhat thanks for sharing LCOO details 16:02:55 np 16:03:11 it looks like we do not have anything else to cover today :) ilyashakhat msimonin ? 16:03:18 a moment :) 16:03:47 ack 16:03:49 DinaBelova, do you have an insight whether you/your colleagues using os-faults? 16:04:10 I was going to make some changes, but it's hard to say who is using it 16:04:15 except me :) 16:04:31 our OpenStack core team should be using it if i'm not mistaken 16:04:46 okey, than I'll be careful 16:05:11 that's all from me :) 16:05:24 ack, thank you sir :) 16:05:35 Can you express link failure in os-faults ? 16:05:50 http://os-faults.readthedocs.io/en/latest/ 16:05:52 * msimonin should read the doc :) 16:06:32 but the purpose is better described in performance-docs 16:06:59 https://docs.openstack.org/developer/performance-docs/test_plans/reliability/version_2/plan.html 16:07:50 ilyashakhat: thanks, I'll go through this 16:09:01 I'm done for today from my side 16:09:14 thank you folks 16:09:17 have a good evening 16:09:24 goodbye! 16:09:27 #endmeeting