00:02:01 #startmeeting congressteammeeting 00:02:02 Meeting started Thu Nov 17 00:02:01 2016 UTC and is due to finish in 60 minutes. The chair is masahito. Information about MeetBot at http://wiki.debian.org/MeetBot. 00:02:03 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 00:02:05 The meeting name has been set to 'congressteammeeting' 00:02:13 hi 00:02:26 ramineni_: hi 00:03:03 today's agenda: 00:03:11 1. PTG 00:03:22 2. status update 00:03:36 anything else? 00:03:45 gate issues? 00:04:19 ramineni_: ok 00:04:21 we could talk about which patches we’re trying to merge for ocata-1 release. 00:04:54 ekcs: got it. 00:05:02 3. gate issue 00:05:09 4. ocata-1 00:05:42 #topic PTG 00:06:27 we needs to decide how many days we have in PTG. 00:06:56 two days or three days 00:07:58 on the mailing list it sounded like the team was fine with 2 days 00:08:02 2 days would be good i guess 00:08:02 it looks like most of us prefer 2 days meeting 00:08:34 +1 (as in 2 days makes sense to me too; not as in 2 + 1 = 3) 00:09:05 * aimeeu thanks ekcs for the explanation 00:10:01 thinrichs also prefers 2 days in the mail 00:10:49 so we choose 2 days session in PTG :-) 00:11:06 great 00:11:13 +1 00:11:28 move on next. 00:11:42 #topic gate issue 00:13:13 ekcs: ramineni_: sorry, I don't catch up recent gate issue. Could you tell us the current issue? 00:13:25 last week we were seeing multiple gate issues - 1.tempest variable deprecation 2. HA timeout 3. doctor tempest 00:13:47 i have submitted one patch https://review.openstack.org/#/c/397013/ 00:14:18 merged it , to actually clear the queue of our review.. hope change sounds good 00:14:55 still we are seeing occasional HA timeout though 00:15:14 thanks ramineni_ 00:15:42 so that patch takes care of the deprecated variable. 00:15:49 can you say a little bit about what the other changes do and why? 00:15:51 ramineni_: thanks. 00:15:59 and HA timeput a lil bit 00:16:34 datasource sync period actually previously its not replaced .. 2 times its added in config file 00:17:52 got it. 00:18:02 so always the latest variable is datasource_sync_period= 30 00:18:16 good catch! 00:18:58 and added synchronize policies in get_status , to get the latest status in first iteration 00:20:07 so still remaining issues are 2 and 3? 00:20:39 masahito: 2 is partially fixed as we are able merge patches now.. 3 is still there 00:21:12 ramineni_: got it. I try to check 3. 00:21:16 i think ekcs raised bug for 3 , may be you could take a look at it 00:22:04 yup. I think I should take care it. 00:22:21 masahito: ekcs: should we consider increasing build timout in gate job? 00:23:06 great work resolving gate so we can merge patches! 00:24:18 ramineni_: I don’t have an opinion on that now. thoughts are: 1. does it happen a lot. 2. can we make things run faster. 00:24:19 ramineni_: any test hits timeout in gate? 00:24:47 and I haven’t looked into it to have knowledge on those questions. 00:25:22 ramineni_ what do you think? 00:25:31 ekcs: masahito: ok, lets see for a week then .. if it happens a lot again, we can think about it then? 00:25:48 ramineni_: Agreed 00:25:54 sounds good. 00:25:56 one more thing: 00:27:57 on the synchronize policy, I wonder if there is another place to trigger the sync instead or another way to do it without having to sync on every get status. 00:29:59 people may expect get status to be a fairly cheap read-only action. syncing all policies may be expensive. 00:30:39 ekcs: I think so. 00:31:11 ekcs: i tried syncing only one policy , but it didnt help HA .. as sometimes its a datasource policy 00:31:43 ekcs: do you have anything on mind on where to place it? 00:32:55 got it. i haven’t thought it all through, but perhaps trigger (rpc) the sync on change (adding/deleting policy; adding/deleting DS), rather than on read. 00:33:58 ekcs: but that doesnt help if multiple peś are there? 00:34:28 ekcs: you mean broadcast? 00:34:37 yea broadcast. 00:34:44 another thought is perhaps we don’t need to sync to memory to report status. maybe we just need to read the policy from DB to report status. with one status being the policy is in DB, but hasn’t been sync’ed to memory. 00:35:30 probably something I need to think through more carefully after ocata-1. 00:36:12 ok, sure, ya Iĺl also think about it 00:36:29 oh and since we’re on the subject, what about this other patch on ha test? 00:36:36 https://review.openstack.org/#/c/397645/ 00:37:23 :) 00:37:47 so primary sync period is left at 30s? 00:37:58 when i see the HA failure , timeout happens in the very initial stage itself, trying ti get status from primary itself 00:38:16 ya, primary sync happens at 30s as we set in devstack 00:38:23 got it. 00:38:41 thanks! 00:38:46 I don’t have anything else on gating. 00:39:24 ok, move on next topic. 00:39:37 #topic ocata-1 release 00:40:23 to release ocata-1 we need to pick patches ready to merge the milestone. 00:40:56 here are the open patches: https://review.openstack.org/#/q/project:openstack/congress+status:open 00:41:24 My opinion is adding patches related to aodh datasource seems to be good. 00:43:22 agreed. maybe each of us with patches open can say which ones we want to merge before ocata-1? 00:43:47 for context, we pretty much need to merge things today to make it for ocata-1. 00:44:32 so essentially things that are ready and maybe just need a quick review. 00:45:08 Sorry I'm late all. Meeting in city ran later than expected 00:45:46 My 2 changes can go in. They've been just waiting on gate. 00:45:50 hi thinrichs ! 00:45:56 thinrichs: hi 00:46:19 ekcs: regarding https://review.openstack.org/#/c/393567/ 00:47:32 any idea why is it still throwing error 00:47:36 ramineni_: i looked into it some more. 00:48:03 the warning saying aodh not found happens only early in the test. I think it’s just because aodh wasn’t ready. 00:49:01 more example in the non-replicated tempest logs, the message doesn’t come up from 16:16 till the end (16:22+) 00:49:19 similar in the replicated pe test. 00:49:44 ekcs: oohk, then should be fine 00:50:54 I think those patches should be mergeable today if looks fine to people. I’ll update it real quick to use info rather than debug as ramineni_ suggested. 00:51:05 ekcs: ok sounds good then 00:51:18 masahito: what about this one? https://review.openstack.org/#/c/397096/ 00:52:52 oh. the bug was assigned to ocata-1. 00:53:42 do you think we can/should merge it today? 00:53:45 I think it's ready to ocata-1. 00:54:09 great! anything else for ocata-1? 00:55:26 #action Let's review these patches and try to merge them today if possible. (Not a big deal if some get delayed and miss ocata-1) 00:55:31 #link https://review.openstack.org/#/c/397096/ 00:55:31 #link https://review.openstack.org/#/c/393420/ 00:55:31 #link https://review.openstack.org/#/c/394258/ 00:55:33 #link https://review.openstack.org/#/c/393977/ 00:55:33 #link https://review.openstack.org/#/c/393567/ 00:55:36 #link https://review.openstack.org/#/c/397796/ 00:55:36 #link https://review.openstack.org/#/c/397096/ 00:56:00 ekcs: thanks the link and wrap up. 00:56:28 last 4 mins. 00:56:44 #topic open discussion 00:57:42 Quick note: I attended the policy meeting this morning, where they're talking about the state of policy.json in OpenStack and whether we could improve it. 00:57:54 Good attendance, over 10 people for sure 00:58:18 thinrichs: sounds big meeting. 00:58:18 Learning about congress is an action item for next week. 00:58:43 awesomeness! 00:58:47 great :) 00:58:49 Discussion of having a backend (Apache fortress or similar) came up. I suggested Congress 00:59:25 Learned something new about policy.json... 01:00:49 oh, running out meeting time. 01:01:42 thanks all. let's review patches, and then good evening/morning/night! 01:02:02 #endmeeting