15:31:31 #startmeeting Performance Team 15:31:32 Meeting started Tue Mar 21 15:31:31 2017 UTC and is due to finish in 60 minutes. The chair is DinaBelova. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:31:33 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:31:36 The meeting name has been set to 'performance_team' 15:31:42 hey folks! 15:31:48 hi o/ 15:31:54 let's see who's around :) 15:32:38 akrzos rook o/ 15:32:54 around :) 15:33:02 :D cool :) 15:33:20 I see none from Inria for now, probably they won't be able to attend 15:33:30 in the meanwhile let's go through action items 15:33:35 #topic Action Items 15:33:42 the first one was on myself 15:34:01 tovin07_ I've contacted Vipin - he's in the middle of relocation to US now 15:34:14 but he'll be working on osprofiler in future as well 15:34:42 yes, good news :D 15:34:43 so he's going to continue being active after some time 15:35:08 tovin07_ one more action item was on you 15:35:09 regarding "evaluate what needs to be done to add result of the function or sql evaluation in the *-stop part of osprofiler trace" 15:35:13 any news here? 15:35:24 Yes 15:35:32 please go ahead 15:35:34 This week, I’m working on that 15:35:55 Plan to push some information later this week 15:36:04 tovin07_ ack 15:36:12 lemme keep this action item for next week as well 15:36:19 Yup 15:36:27 #action tovin07_ evaluate what needs to be done to add result of the function or sql evaluation in the *-stop part of osprofiler trace 15:36:36 also one more action item was on msimonin 15:36:39 rcherrueau: o/ 15:36:42 o/ 15:36:43 but I do not see him here 15:36:48 rcherrueau o/ 15:37:08 rcherrueau don't you know status of msimonin action item? 15:37:09 that was about "ping Rally PTL regarding adding osprofiler support to rally code base" 15:37:41 I don't think he did it, sorry about that 15:37:46 it's ok 15:37:52 lemme keep it for the next week as well 15:37:59 #action msimonin ping Rally PTL regarding adding osprofiler support to rally code base 15:38:16 ok. so we're done with action items 15:38:23 #topic Current progress on the planned tests 15:38:28 rcherrueau any news? 15:38:52 Yes one 15:39:31 msimonin did an OpenStack provider for enos, the tool we develop for measuring OpenStack 15:39:46 it means we can deploy an OpenStack over an OpenStack 15:40:03 o/ 15:40:34 rcherrueau just curious - for what use case specifically? I guess there might be lots of reasons but still :) 15:40:37 and then measure it. He did it because we are currently working on Chameleon testbed 15:41:09 rcherrueau ok, gotcha 15:41:10 Chameleon is a testbed provided by university of Chicago 15:41:32 and Chameleon is operated though an OpenStack 15:41:44 through* 15:42:17 that's all for our side 15:42:23 ack, thanks 15:43:09 rook any updates from RedHat? :) we all are curious if akrzos was able to puch Gnocchi margins further :) 15:44:18 it looks like rook went afk for a while :) 15:44:24 DinaBelova: I have been head down doing a scale deployment in CNCF 15:44:33 rook wow 15:45:01 DinaBelova: wrt to Gnocchi 15:45:08 what scale are we talking about? 15:45:13 I'm working with Ocata bits and intend on getting another test ruig setup 15:45:23 s/ruig/rig/ 15:45:33 in the near future 15:45:36 before Summit 15:45:37 DinaBelova Install and OpenStack scale -- and OpenShift on OpenStack scale 15:45:39 akrzos ack, thank you sir 15:45:52 rook wow, huge task 15:45:53 i have a question though 15:45:56 wrt to Rally 15:46:05 Gnocchi is moving out of big tent 15:46:12 rook good luck man :) those tasks are usually really tricky :) 15:46:20 akrzos oh, interesting 15:46:24 did not know about this 15:46:25 will it still be able to sit inside rally or will rally no longer want it if it's not in OpenStack Big Tent? 15:47:09 akrzos please ping andreykurilin about this 15:47:15 akrzos in #openstack-rally channel 15:47:55 akrzos btw, don't you know the details? 15:47:55 ok will do 15:48:05 why gnocchi is moving out from BT? 15:48:35 they are looking toget more contributors and it's not glued to OpenStack 15:48:48 meaning you can deploy it outside of OpenStack 15:48:50 Oleg Basov proposed openstack/performance-docs master: Influx HA as an external storage for Prometheus https://review.openstack.org/442020 15:49:05 let me fine the mail message 15:49:23 akrzos ok, thanks - it looks like I missed the announcement about this 15:50:17 #link http://lists.openstack.org/pipermail/openstack-dev/2017-March/114300.html 15:50:23 akrzos thanks 15:50:36 np 15:51:13 ok, from mirantis side - no many updates from me today. we still have all k8s and containerized openstack related activities in progress. this is going to be a long process anyway :D so we're working through the deployment issues and test cases for now 15:51:39 and frankly speaking that's all from my side for now 15:52:05 it looks like we may jump to open discussion 15:52:06 #topic Open Discussion 15:52:14 how is the performance of containerized services? 15:52:19 big differences? 15:53:08 rook I won't say so 15:53:14 sometimes we observed issues, but they were related more to the networking setup 15:53:37 the virt networking? 15:53:53 or do you guys pass the networking stack to the container? 15:54:19 from OSprofiler, patches in Zun were merged last week 15:54:19 I’m continue to work with other patches 15:54:27 sometimes we observed slow container creation and scheduling on scale, but that's more about k8s and docker, it needs to be cooked right 15:54:33 rook indeed 15:54:44 we're experimenting with Calico networking mostly 15:55:09 due to the fact that was a request from our k8s team 15:55:51 rook and it's having several setup options that are use-case specific, OpenStack is indeed very specific applicaiotn 15:55:59 rook: @inria, we have a PhD student on that specific topic. Her first work was on VM, in particular, the impact of I/O with and without virtio. She recently start a same job with container. Hope I can link to her conclusion soon. 15:56:01 tovin07_ thanks 15:56:29 rcherrueau that is going to be a good research, are you planning to share it in community? 15:57:15 Yes, of course, but she has, first, to publish in a scientific conference. 15:57:53 rcherrueau cool, please give us links once they will be available :) 15:58:07 I will do 15:58:14 thanks 15:58:27 anything else to cover today? we had fast meeting :) 15:58:32 besides, DinaBelova, did you discuss with vipin about proposing me as core contributor in OSprofiler?’ 15:58:49 tovin07_ sadly nope, Vipin is not that available those days due to the relocation 15:59:09 will make sure it'll happen this week anyway 15:59:14 tovin07_ don't worry :) 15:59:26 seems like he’s very busy 15:59:37 DinaBelova: There will be a dedicated slot for performance WG at Boston? 15:59:41 Okay, thanks :D 16:00:08 rcherrueau it's not going to be a usual summit where some design sessions were held 16:00:27 so that's going to be only main, presentation / market place part of the summit 16:00:27 DinaBelova: maybe a Forum 16:00:39 rcherrueau nope, for now it's not planned 16:00:52 DinaBelova: Ok thanks 16:01:09 although we still can hand somewhere to discuss most recent stuff in person anyway 16:01:15 hang* 16:01:33 it looks like we're done for today 16:01:45 thanks everyone for participating 16:01:50 thanks 16:01:54 #endmeeting