16:00:01 <DinaBelova> #startmeeting Performance Team
16:00:02 <openstack> Meeting started Tue Feb  2 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.
16:00:03 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
16:00:05 <openstack> The meeting name has been set to 'performance_team'
16:00:14 <DinaBelova> hello folks :)
16:00:22 <DinaBelova> I know regXboi and rvasilets__ is here :)
16:00:38 <DinaBelova> kun_huang won't attend this particular meeting
16:00:50 <DinaBelova> Dunno if Boris will be around
16:01:04 <regXboi> yes, but I'm also on phone calls in this hour :( so I don't know how much I'll be paying attention
16:01:04 <DinaBelova> harlowja - are you here? :)
16:01:09 <DinaBelova> regXboi ack
16:01:32 <DinaBelova> ik, so let's start with the action items
16:01:33 <DinaBelova> #topic Action Items
16:01:47 <DinaBelova> I can say that last week was busy enough for me and kun_huang
16:02:01 <DinaBelova> therefore no lab info published yet from both of us
16:02:10 <DinaBelova> we're really sad, folks, to say that
16:02:23 <DinaBelova> kun_huang pinged DragonFlow team
16:02:44 <DinaBelova> they are working on performance data plane testing plan for now
16:02:46 <DinaBelova> #link https://review.openstack.org/#/c/272559/
16:02:53 <DinaBelova> #action DinaBelova review https://review.openstack.org/#/c/272559/
16:03:34 <DinaBelova> I suppose the unified test plan will be moved to the performance-docs once it'll be more or less ready
16:03:51 <DinaBelova> in fact that's all news about the action items, sorry folks for not too much ipdate here
16:03:54 <DinaBelova> update*
16:04:18 <DinaBelova> a bit more info about test plans published to performance-docs
16:04:20 <DinaBelova> #topic Test plans statuses
16:04:57 <DinaBelova> during last week ilyashakhat has proposed https://review.openstack.org/#/c/270204/
16:05:08 <DinaBelova> I see Leontiy did not manage to review it
16:05:28 <DinaBelova> lemme become a despot a bit and merge it by myself
16:06:03 <DinaBelova> kun_huang has proposed the cool idea about splitting the test plans to something abstract + the example
16:06:10 <DinaBelova> #link https://review.openstack.org/#/c/275092/2
16:06:22 <DinaBelova> so one more patch set to review now
16:06:52 <DinaBelova> ok, cool
16:07:07 <DinaBelova> any questions regarding the test plans?
16:07:20 <openstackgerrit> Merged openstack/performance-docs: Add test plan for tenant networking  https://review.openstack.org/270204
16:07:50 <DinaBelova> ok, it looks like no questions yet :)
16:08:23 <DinaBelova> I'll skip Huawei lab topic as kun_huang is absent today
16:08:41 <DinaBelova> #topic OSProfiler weekly update
16:08:56 <DinaBelova> ok, so during last week three changes were merged
16:09:10 <DinaBelova> #link https://review.openstack.org/269860
16:09:18 <DinaBelova> #link https://review.openstack.org/270898
16:09:25 <DinaBelova> #link https://review.openstack.org/268974
16:09:43 <DinaBelova> the most important was he first in this list, that was blocker for the release
16:10:01 <DinaBelova> so today Boris and I will cut new OSprofiler release
16:10:20 <DinaBelova> that means that today I'll update the nova change
16:10:26 <DinaBelova> #link https://review.openstack.org/254703
16:10:31 <DinaBelova> neutron change
16:10:38 <DinaBelova> #link https://review.openstack.org/273951
16:10:42 <DinaBelova> and keystone one
16:10:50 <DinaBelova> #link https://review.openstack.org/103368
16:10:51 <regXboi> yes, that should help the UT failures on the neutron PS
16:11:01 <DinaBelova> regXboi - yep, indeed
16:11:05 <DinaBelova> johnthetubaguy ^^
16:11:20 <DinaBelova> I believe new nova change patch set will appear today
16:11:57 <DinaBelova> and as this change is almost final, I believe that if you'll be able to review it, we have good changes to land it in Mitaka at the end
16:12:56 <DinaBelova> neutron and keystone changes are not that much tighten to the time, as they are 'bugs fixes', nova change is pushed as a 'small specless feature'
16:13:06 <DinaBelova> so thanks again Nova folks for being patient
16:13:41 <DinaBelova> in fact these ^^ changes are most important for now
16:14:01 <DinaBelova> so I suppose I'll focus on them today and tomorrow
16:14:38 <DinaBelova> regXboi - I've generated the primary trace for the nova boot with neutron
16:14:48 <regXboi> cool
16:14:56 <DinaBelova> #link http://dinabelova.github.io/nova_boot_with_neutron.html
16:15:13 <DinaBelova> regXboi - I'm still trying to understand what I miss there :)
16:15:30 * regXboi looks
16:15:37 <DinaBelova> the only issue with nova boot tracing in case of OVS agent used is the following
16:16:18 <DinaBelova> as neutron-server creates the port for the VM without agent participating, OVS agent find out new port on OVS is placed during periodical polls
16:16:40 <regXboi> the 25 second rpc to neutron-ovs-agent is ... well ... scary
16:17:15 <DinaBelova> regXboi - heh, yeah, but it's not the time then ack comes back
16:17:23 <DinaBelova> but the time actual operations happen
16:17:51 <regXboi> uh, doesn't that imply a race condition?
16:18:28 <DinaBelova> regXboi - dunno yet, need to investigate
16:18:38 <regXboi> ack
16:18:48 <DinaBelova> I may be wrong in fact with my "ack-no ack" assumption
16:19:04 <DinaBelova> that's - let's say - my thoughts about what I see
16:19:20 <regXboi> well... looking at it, that is the thing that worries me the most
16:19:41 <regXboi> so if we nail that down, then we can walk back through the rest of the long poles
16:19:51 <DinaBelova> also we need to keep in mind that security groups operations are heavy and on devstack all-in-one VM they may be crazy anough
16:19:56 <DinaBelova> regXboi ack
16:20:17 <DinaBelova> ok, cool
16:20:47 <DinaBelova> so today I'll continue with this ^^ security group operation and with trying to trace l3 agent
16:21:25 <DinaBelova> regXboi - btw, sadly I'm not sure if I generated this trace without debugs, so we may face this too long time because of stupid pdb somewhere :)
16:21:25 <regXboi> once we have the new osprofiler cut and requirements updated, I can re-check the neutron patch locally
16:21:30 <DinaBelova> I'll need to regenerate
16:21:36 <DinaBelova> regXboi ack
16:21:50 <regXboi> DinaBelova - it is what it is - that's part of the fun
16:22:01 <DinaBelova> regXboi :D
16:22:19 <DinaBelova> ok, so that's almost everything about the osprofiler for now
16:22:31 <DinaBelova> any quesitons?
16:23:03 <DinaBelova> ok, cool
16:23:05 <DinaBelova> #topic Open Discussion
16:23:31 <DinaBelova> in fact because of no Boris and Kun attending this particular meeting we did not have much to discuss today:)
16:24:10 <regXboi> nothing wrong with getting some time back
16:24:31 <DinaBelova> regXboi that's true :)
16:24:41 <DinaBelova> last time I've added osprofiler-specific topics to opw next round topics
16:24:55 <DinaBelova> no updates from that field yet, so I'll keep you folks tuned
16:25:22 <DinaBelova> also I've proposed two osprofiler-specific talks on summit - will send links to everyone once voting will be started
16:25:31 <DinaBelova> thanks everyone for participating!
16:25:40 <DinaBelova> bye!
16:25:51 <DinaBelova> #endmeeting