16:00:01 #startmeeting Performance Team 16:00:01 Meeting started Tue Mar 22 16:00:01 2016 UTC and is due to finish in 60 minutes. The chair is DinaBelova. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:02 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:00:04 The meeting name has been set to 'performance_team' 16:00:09 o/ 16:00:13 o/ 16:00:34 let's wait for a few moments to see if others will be here 16:00:58 \o 16:01:30 rohanion decided to be not like others :) 16:02:24 just saying 'hi' 16:02:41 :) 16:02:51 let's see if ilyashakhat and listomin will join 16:04:43 ok, so let's start and if somebody will join in the meanwhile I'll be glad :) 16:05:00 today we have standard agenda with action items first 16:05:12 #topic Action Items 16:05:39 I did not collected enough feedback to separated meeting to two time zones - US and European friendly 16:06:03 it looks like after moving to daylight saving time in US everyone became happy with current time 16:06:21 so if no complains will come I'll leave this time for a meeting 16:06:30 rohanion - what's about your items? 16:06:40 #link http://eavesdrop.openstack.org/meetings/performance_team/2016/performance_team.2016-03-15-16.00.html 16:06:48 #1 collect data about possible performance degradation of light weighted proxies for 0MQ 16:06:56 #2 prepare list of possible drivers for osprofiler with pros/cons of each one 16:07:28 first one - unfortunately nothing, I've been working on osprofiler the whole week 16:08:05 second - almost done with looking for the most inportand driver https://etherpad.openstack.org/p/osprofiler-engines 16:08:09 important* 16:08:40 #action everyone take a look on https://etherpad.openstack.org/p/osprofiler-engines and add comments about priority, etc. 16:08:50 #link https://etherpad.openstack.org/p/osprofiler-engines 16:09:30 as for Ilya's action item it's still in progress 16:09:31 #link https://review.openstack.org/#/c/291657/ 16:09:52 due to the fact change that will create separated repo for performa is not yet merged to the openstack-infra/project-config 16:10:03 ilyashakhat move Performa to openstack space (final steps) 16:10:07 #action ilyashakhat move Performa to openstack space (final steps) 16:10:16 so let's keep this action item so far 16:10:23 hi! 16:10:28 i made it 16:10:28 harlowja_at_home wooooow 16:10:29 lol 16:10:34 it's so unusual 16:10:35 ЖВ 16:10:38 lol 16:10:38 :D ** 16:10:43 its like a bday present 16:10:46 josh the bday present 16:10:48 lol 16:10:52 :D 16:11:00 ok, very interesting :) 16:11:24 an idea, btw, do u guys want to start having some of the osprofiler stuff in the oslo weekly meeting to? 16:11:31 I do not see kun_huang today (and in fact I did not see him for a whole week... so let's skip his action item for now) 16:11:45 since i just saw https://etherpad.openstack.org/p/osprofiler-engines 16:11:45 harlowja_at_home probably this will be a good idea 16:11:53 DinaBelova, or at least the osprofiler stuff there? 16:11:55 just an idea 16:12:06 guess i'm PTL of oslo or something, lol 16:12:11 harlowja_at_home btw when is the oslo team meeting? 16:12:15 harlowja_at_home yeah, I know 16:12:21 congrats, man, btw! 16:12:24 :-P 16:12:26 thx 16:12:36 http://eavesdrop.openstack.org/#Oslo_Team_Meeting DinaBelova 16:12:41 ack 16:12:43 good job, congrats! 16:12:45 lol 16:12:54 thx 16:12:59 it's nice time :) Europe friendly 16:13:06 :-P 16:13:08 cool, will add to my calendar 16:13:26 aren't u supposed to come back to the states sometime? 16:13:26 lol 16:13:43 harlowja_at_home hahah, I need to survive till September somehow anyway 16:13:57 :) 16:14:12 ok, so going to the test plans statuses 16:14:13 #topic Test plans statuses 16:14:32 I still do not see ilyashkhat, so lemme tell what I know for him 16:14:40 #link https://review.openstack.org/#/c/295919/ 16:15:10 here is current status of the MQ HA test report 16:15:20 that he performed using performa :D 16:15:37 change is really fresh, so I did not have a chance to look at it - will do this week 16:15:51 and I see listomin has joined 16:16:10 Yeah, Hello! 16:16:35 Here is the test plan and results for docker repositories services https://review.openstack.org/#/c/295308/ 16:16:47 can you share details about your containers repos testing? that will be useful for people who would like to run containerized OpenStack 16:16:54 #link https://review.openstack.org/#/c/295308/ 16:17:17 still doubt if we do need all *.csv files in raw format 16:17:24 0_0 16:17:40 this makes change a bit monstrous 16:17:45 listomin what do you think? 16:18:38 I believe we need to keep all CSVs because others can use it to perform some calculations. Probably we need to store it in other format 16:19:37 As for test results. We have found that we can upload only 5 images at the same time 16:19:45 let's probably split the whole change to two: the result+plan and the raw data - and let's rethink where to store this raw data 16:19:52 It seems it's limitation in docker 1.10 16:19:54 listomin to all of the repos? 16:20:18 yep, for each of the repo services 16:20:26 ack, it's interesting 16:20:40 please keep us updated if that's docker 1.10 or something else 16:20:51 if you'll find something out 16:21:12 Sure, I've created a place for this note 16:21:19 in results 16:22:00 ok, so on my side I'm playing with Keystone DB/Cache usage effectiveness - now I have draft result from which I can say that sqlalchemy usage was indeed improved in Mitaka keystone 16:22:17 I did not manage to create format test plan yet, going to do it this week 16:22:55 #action DinaBelova add keystone testing plan/first results to the performance-docs 16:23:19 anything else to cover here? 16:24:00 ok, let's proceed 16:24:01 #topic OSProfiler weekly update 16:24:30 there was bunch of small changes this week 16:24:34 related to keystone profiling 16:24:51 #info https://review.openstack.org/294515 - Clean thread local profiler object after usage 16:25:01 #info https://review.openstack.org/294516 - Add hepler to trace sessions inside _TransactionContextManager 16:25:15 #info https://review.openstack.org/294533 - Add possibility to trace _TransactionContext resource 16:25:23 #info https://review.openstack.org/294535 - Add DB operations tracing 16:25:40 so in fact right now we have argues about #2 and #3 16:25:53 in particular where to place this code 16:26:09 oslo.db / keystone / osprofiler are possible options for now 16:26:18 harlowja_at_home btw please review https://review.openstack.org/294515 16:26:22 very funny issue 16:26:42 yes ma'm 16:27:00 harlowja_at_home can you imagine my face when I saw DB operation in the trace on the upper level without any parents in keystone tracing 16:27:01 :D 16:27:10 lol 16:27:13 i can imagine 16:27:36 but i don't want to imagine for to long 16:27:38 lol 16:27:40 :D 16:27:45 ahaha, that's true 16:28:00 so it's 5 lines change that makes game different 16:28:20 we did not face it for other projects as they are using greenlets almost everywhere 16:28:33 and keystone uses reusable threads pool :) 16:28:50 rohanion - any updates on drivers side? 16:31:36 well, none yet 16:32:04 rohanion ok, so I suppose you spent time on playing with profiler 16:32:18 but I'm familiar with osprofiler now and I want to merge Dina's commits asap 16:32:26 DinaBelova, ok, comment added on that review 16:32:38 merge all my commits asap to :-P 16:32:42 (in other projects, ha) 16:32:47 harlowja_at_home ack - will do 16:32:50 :D 16:33:13 k, so it looks like that's it for profiler update 16:33:24 let's go to the open discussion 16:33:24 #topic Open Discussion 16:34:00 i will finish what you've started 16:34:03 http://images.fromupnorth.com/355/566ed31651b13.jpg 16:34:21 damn DinaBelova u look differerent in that picture 16:34:22 lol 16:34:24 :) 16:34:38 harlowja_at_home well you do bot usually see me without makeup :D 16:34:43 lol 16:35:01 we do not have our time slot defined yet for summit 16:35:31 although working group sesison was approved, design summit slots are not yet scheduled 16:35:40 did u guys want someting in oslo? 16:35:42 https://etherpad.openstack.org/p/newton-oslo-summit-planning 16:35:44 so I'll keep you guys updated 16:36:02 i can try to get one for osprofiler (design summit thing) 16:36:08 whenever i get to schedule things 16:36:23 harlowja_at_home - ack, but I had my proposal already approved 16:36:26 k 16:36:34 wellllll let me know 16:36:34 ha 16:36:34 so I think one session will be enough 16:36:39 harlowja_at_home sure :F 16:36:42 :D 16:37:03 harlowja_at_home thanks for the link to oslo newton plans, I'll need to go through it 16:37:12 k 16:37:12 np 16:37:21 ok, so anything else to cover? 16:37:48 ok, cool 16:38:04 so thanks everyone (especially harlowja_at_home :D) for attending 16:38:09 #endmeeting