16:00:06 #startmeeting Performance Team 16:00:08 Meeting started Tue Jun 7 16:00:06 2016 UTC and is due to finish in 60 minutes. The chair is DinaBelova. Information about MeetBot at http://wiki.debian.org/MeetBot. 16:00:09 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 16:00:11 The meeting name has been set to 'performance_team' 16:00:19 hey everyone :) 16:00:51 who's here? :) 16:00:57 here :) 16:01:06 hi o/ 16:01:10 Hello 16:01:10 rohanion, ad_rien_ ? 16:01:17 Sorry I cannot stay 16:01:19 meow? 16:01:21 today 16:01:25 wow, it's much quicker start today 16:01:53 ad_rien_ np sir, I just want to say sorry I did not have enough time to work on the my action items regarding researches we both are interested in 16:02:22 sure as I told you, as a researcher institute we are also progressing slowly 16:02:23 ad_rien_ I mean http://execo.gforge.inria.fr/doc/latest-stable/ and checking current status of internal Mirantis JJ and python scripts - if they can be made public right now 16:02:24 here =) 16:02:29 ad_rien_ ack 16:02:32 so there is no problem 16:02:38 anevenchannyy msimonin1 o/ 16:02:40 gokrokve o/ 16:02:40 o/ 16:02:47 ok, so let's get started 16:02:51 #topic Action Items 16:03:14 as said, sadly I'm very busy right now with internal Mirantis product release cycle 16:03:29 so was not able to work 100% on my tasks 16:03:39 although I jave some news regarding Ironic tests/results publishing 16:03:52 in fact tests most probably will be run in July 16:03:57 o/ 16:04:08 rook o/ 16:04:21 some preliminary results were sent to the community, btw, some time ago 16:04:37 #link http://lists.openstack.org/pipermail/openstack-dev/2016-April/091933.html 16:04:56 so folks are going to start with this info publishing and then continue deeper analysis 16:05:29 rook - did you have a chance to speak with ilyashakhat before his vacation? 16:05:39 regarding the offloading model suggestion 16:06:08 in the meanwhile leaving action items important for ad_rien_ till next time 16:06:14 #action DinaBelova go through the http://execo.gforge.inria.fr/doc/latest-stable/ and compare it with the approach used inside Mirantis 16:06:23 #action DinaBelova check current status of internal Mirantis JJ and python scripts - if they can be made public right now. Evaluate what needs to be done otherwise 16:06:30 rook ? :) 16:07:07 DinaBelova: no, i was on vacation the past 2 weeks ;) 16:07:12 rook ahaha :D 16:07:15 so we didn't sync up :) 16:07:26 so ilyashakhat will be back in a week :) 16:07:40 your vacations are overlapping for 1 week :) 16:07:48 ok, so let's leave this action item as well 16:07:51 omg all our team went on vacation 16:07:55 haha funny 16:08:01 #action rook ilyashakhat go through the offloading model suggestion 16:08:32 rook - just for future - may you please tell me next time once you'll plan to miss the meeting? Sometimes there is a great wish to discuss something with you 16:08:47 and it's a bit confusing when I don't know if you're around :) 16:09:26 it looks like that's all for action items progress for now 16:09:38 let's jump to the test plancs 16:09:43 plans* 16:09:46 #topic Current progress on the planned tests 16:10:12 so right now we've finished some preliminary Openstack cloud check on the scale env 16:10:19 DinaBelova: ah ha, sure! you can also reach me via email too DinaBelova (rook@redhat.com) 16:10:22 and are going to run baseline tests 16:10:27 rook sure :) 16:10:54 after that reliability tests are planned - and we're having draft test plan published on review today 16:11:01 #link https://review.openstack.org/326598 16:11:16 folks, I'm really encouraging you guys to review it and leave your comments 16:11:37 also we have some test plans and result coming from Mirantis Neutron team 16:11:49 here is the second one 16:11:53 #link https://review.openstack.org/324596 16:12:02 just a reminder - the first one was this 16:12:14 #link https://review.openstack.org/310419 16:12:41 also alwex (interesting that he's not around) is working on 1000 nodes test plan 16:12:42 #link https://review.openstack.org/324806 16:13:00 some comments need to be fixed 16:13:14 gokrokve - may you please tell alwex to take a look on mine ones as well? 16:13:31 sure 16:13:36 He will do this otday 16:13:58 ack, cool 16:14:23 so right now (until baseline results will be collected) I have nothing more to share 16:14:33 anything from lezbar and ad_rien_ ? 16:15:20 not so far 16:15:20 Hello I'm speaking for ad_rien 16:15:26 msimonin - sure 16:16:04 We'll look at these test plans 16:16:21 yes, please (!) share your ideas and comments 16:16:24 this is really appreciated 16:16:27 If needed we can run the same experiments on our platforms 16:17:11 especially the 1000 nodes test 16:17:16 i think we can (at least) do that for tenant networking and mq layer + 1000 nodes test 16:17:24 to have comparison between the envs 16:17:45 I guess this will be super useful - as right now we're having only our lab researches 16:18:05 Do we have access to the deployment script (sorry I just go through the document very quickly) ? 16:18:36 msimonin - are you talking about 1000 nodes test plan? 16:18:47 DinaBelova: yes 16:19:28 i think alwex did not share it outside the mirantis so far... gokrokve - may you please encourage alwex to join the IRC and then speak with msimonin ? 16:19:43 ok 16:19:55 probably right after a meeting or whenever 16:20:05 msimonin - alwex is PST time zone 16:20:08 he is not in the office yet 16:20:29 if he'll be too late today, just let's move this discussion to tomorrow 16:20:32 gokrokve ack 16:20:55 From our side we'll have more time during the summer to perform the evaluation 16:21:27 msimonin ack, may you please share a bit more specific schedule ? 16:22:04 probably next time? 16:23:17 DinaBelova: i can give you a precise answer next time, but I think having the evaluation by September is doable 16:23:44 #action msimonin share some details on evaluation schedule possible to be done by Inria 16:23:48 ack, thank you sir 16:23:59 it looks like we may proceed 16:24:10 #topic OSProfiler weekly update 16:24:16 hey rohanion o/ 16:24:21 hiii 16:24:24 I know you're just after the vacation 16:24:27 but any news? 16:24:46 yeah, I've got something 16:25:14 We're continuing working on integration of osp into horizon ui 16:26:00 so basically all the functionality in the core driver structure and messaging/ceilometer/mongodb is implemented 16:26:12 and what's about the fixing -1 comments? :) 16:26:28 I will do that in a week I hope. 16:26:47 rohanion ack, thanks 16:26:48 It's more important to deliver osprofiler to horizon 16:26:51 for now 16:27:16 rohanion - just an ask for you sir - to spend some time on the perofrmance-docs reviews as well 16:27:30 I think your feedback will be valuable 16:27:58 ok, it looks like we'll have more osprofiler specific updates next time 16:28:02 #topic Open Discussion 16:28:07 after we create something that really works well and horizon will greet us with a shiny osprofiler base_id trees, I will return to fixing all the -1s 16:28:18 DinaBelova: I saw there were some preliminary results published from the 1000 node test, any analysis done? 16:28:36 bsilverman it's the preliminary test plan in fact 16:29:01 and some very primary result 16:29:19 what we had - we ran 1000 compute nodes using containers 16:29:29 against really small lab as a PoC 16:29:44 the basic rally scenario was used to check the workability 16:29:53 with noop instances being ran 16:30:23 the next step is to extend the test and run it against bigger HW lab and run more realistic scenaios 16:30:39 to see how various control plane services will feel on such setup 16:30:51 this is something to be done 16:31:01 Okay, do you have a date for when the lab will be ready? 16:31:28 it's gonna happen after baseline, load and reliability tsting being passed 16:31:36 so not in the next 2 weeks 16:31:44 but hopefully at the end of June 16:32:13 okay 16:32:16 bsilverman ^^ this is current plan 16:32:17 ok 16:32:26 Who is doing the test plan management now? 16:33:25 inside mirantis we're having some plans for Q2 and trying to reach them - like baseline, load, density, reliability and 1000 nodes emulation 16:33:36 so this is for this month 16:33:54 DinaBelova: ack 16:34:03 let's see what we'll be able to run in fact and we'll manage schedule accordingly 16:34:36 as for forks from Inria - they are planning to join (afair, msimonin please corrent) actively closer to the middle of summer 16:35:00 DinaBelova: yes exactly 16:35:06 ack, cool 16:35:24 ok, anything else to cover as a part of open discussion? 16:35:29 gokrokve, rook ? 16:35:44 nothing from my side 16:35:45 I'll try to catch alwex in the mean time :) 16:35:48 nada 16:36:06 msimonin ack 16:36:15 ok, folks, it looks we were super quick today 16:36:21 thanks everyone for coming 16:36:26 have a nice day/evening 16:36:32 #endmeeting