00:01:06 <ekcs> #startmeeting CongressTeamMeeting 00:01:07 <openstack> Meeting started Thu Aug 4 00:01:06 2016 UTC and is due to finish in 60 minutes. The chair is ekcs. Information about MeetBot at http://wiki.debian.org/MeetBot. 00:01:08 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 00:01:10 <openstack> The meeting name has been set to 'congressteammeeting' 00:01:13 <ramineni_> hi 00:01:16 <aimeeu> Hello 00:01:41 <ekcs> Hi all! Hope everyone's having a good week. 00:01:51 <ekcs> thinrichs is away this week and asked me to fill in on the meeting. 00:02:45 <ekcs> looks like tsandall is with us today. Hi! 00:03:12 <ekcs> Topics for today: 00:03:12 <ekcs> 1. Gating 00:03:13 <ekcs> 2. Status updates 00:03:14 <ekcs> Any other topics? 00:05:00 <ekcs> #topic gating 00:05:13 <ekcs> Big thanks for ramineni_ and others who fixed the recent gating issues. 00:05:57 <ekcs> It seems to have fixed the intermittant gate-congress-new-arch-devstack failures. Only exception is when I tried to recheck but forgot to rebase a patch. 00:05:57 <ekcs> Anything else on the topic? 00:06:36 <ramineni_> ekcs: with your locking patch, hope we wont have any more gating issues 00:07:05 <ekcs> hopefully =) 00:07:11 <ramineni_> it should take care of any further race conditions right 00:07:14 <ramineni_> :) 00:07:29 <masahito> sorry, late. 00:07:34 <ekcs> hi masahito 00:07:53 <masahito> ekcs: hi 00:08:02 <ekcs> masahito, we’re just talking about the recent gating issue and that we seem to have calmed it down. 00:08:25 <masahito> got it. 00:08:39 <ekcs> moving on then. 00:08:41 <ekcs> #topic status 00:08:56 <masahito> I think ramineni_ did great works for it. 00:09:15 <ekcs> masahito: yup! 00:09:24 <ramineni_> masahito:ekcs: thanks :) 00:09:37 <ekcs> ramineni_: would you like to start us off on status updates? 00:09:55 <ramineni_> ekcs: sure 00:11:14 <ramineni_> ekcs: this week mainly worked on gating issues, and made python3.5 job voting as it seems to be passing now, 00:11:35 <ramineni_> ekcs: planning to add patch on follow up of Timś to remove old dse code 00:11:58 <ramineni_> ekcs: thats it from my side 00:12:24 <ekcs> great! Great work getting on top of all that. 00:13:14 <ekcs> yea it may be a good idea to follow up on remove old DSE before it becomes terribly out of date. Thanks! 00:13:29 <ekcs> aimeeu: would you like to go next? 00:13:33 <aimeeu> sure 00:14:02 <aimeeu> I finally pushed the first version of the HA Overview Guide and deployment instructions. 00:14:02 <aimeeu> https://review.openstack.org/#/c/350731/1 I marked the sections I'd like help with 00:14:43 <aimeeu> For the other minor bug fixes, I was waiting for the Horizon plugin model patch to merge 00:14:48 <aimeeu> #link https://review.openstack.org/#/c/305063/ 00:15:15 <aimeeu> so I don't end up in git/gerrit hell 00:15:29 <aimeeu> That's it for me 00:17:02 <ekcs> thanks aimeeu! can you remind us which horizon plugin model patch we’re waiting on? 00:17:06 <aimeeu> I welcome any and all comments on the overview and deployment guides. I did a lot of reading but have no experience with HA 00:17:19 <aimeeu> ekcs: https://review.openstack.org/#/c/305063/ 00:17:35 <aimeeu> Anusha just went in forced a recheck 00:17:58 <ekcs> aimeeu: ah got it. 00:18:32 <aimeeu> Hopefully this coming week I will have more time to work on Congress. 00:19:08 <ekcs> and great work jumping on the HA. let’s all take a quick look to see if there are things we can help aimeeu with on the HA docs. 00:19:51 <ekcs> I’ll help with the section on action execution during failover. 00:20:08 <ekcs> #action ekcs help with action execution failover in HA docs. 00:20:15 <ekcs> thanks aimeeu! 00:20:21 <masahito> ok, I'll check ha-deployment section. 00:20:24 <aimeeu> ekcs: excellent. Many thanks. It's always easier to edit or add, and since I copied a lot from your blueprint... 00:20:54 <aimeeu> masahito: thanks - I added the active-active paragraph to your ha-deployment guide 00:21:27 <ekcs> awesome. 00:21:33 <ekcs> masahito: you wanna go next? and feel free to add topics to the agenda since we missed you during that section. 00:21:50 <masahito> ok 00:21:53 <ekcs> actually sorry 00:22:10 <masahito> I'm working for lazy datasource feature. I pushed first patch that is ready to review and will continue to implement the feature to all datasources. 00:22:23 <ekcs> nvm go ahead =p 00:22:29 <masahito> that's from my side. 00:22:58 <ekcs> thanks masahito! 00:22:59 <masahito> sorry, I didn't go into HAHT stuff this week. 00:23:16 <ekcs> masahito no problem! 00:23:45 <ekcs> masahito: one question about lazy datasource, 00:24:00 <masahito> ok. 00:24:14 <ekcs> here’s the link btw: #link https://review.openstack.org/#/c/346677/ 00:25:35 <ekcs> I was having a little trouble understanding whether the same update method would get run multiple times on some existing DSDs not yet adapted specifically for lazy. 00:27:49 <masahito> I expect only tables specified as 'lazy_table' don't pull data from real datasource. 00:29:06 <ekcs> ok thanks. I think I need to understand it more. Will do that this week. 00:29:44 <masahito> ekcs: thanks. I wrote down how to use the feature in commit message. 00:30:01 <ekcs> thanks. 00:30:13 <ekcs> Ok here's my status 00:30:14 <ekcs> The persist push data patch is under review. https://review.openstack.org/#/c/348712/ 00:30:30 <ekcs> I think we're ready to do more integration testing on HA, which I'll start this week. 00:31:17 <ekcs> Thanks for the review on that masahito 00:31:19 <ekcs> Another small change to trigger synchronize_policies after adding datasource, and adding a lock to avoid concurrency errors. https://review.openstack.org/#/c/350811/ 00:31:51 <ekcs> that’s all for me. 00:33:30 <ekcs> Great let's move on to open discussion then. 00:33:33 <ekcs> #topic open discussion 00:34:32 <masahito> Speaking of HAHT, do we want to have another jenkins job for multi-node deployment like gate-congress-dsvm-multi-node-deployment? or implement tempest tests? 00:36:36 <ramineni_> masahito: ya, i think we should do 00:36:47 <ekcs> masahito: good question. 00:36:52 <ekcs> Any thoughts on pros and cons either way? 00:38:57 <masahito> new job Pros - we can make different tests senario for HA from cold-stanby HA, Cons - we need to implement the test from scratch. 00:39:37 <ekcs> masahito: got it. 00:40:21 <ekcs> do we think we can test the setups we need by adding to existing HA tests? 00:40:49 <masahito> new tempest test Pros - we can re-use tempest scenarios we already have, Cons - we sometimes struggle from some errors. 00:41:27 <ekcs> got it. 00:41:38 <ekcs> ramineni_: do you have any thoughts? 00:42:04 <masahito> I think the existing HA test expects cold-stanby HA. 00:42:25 <ekcs> I guess there are a few options: 00:42:32 <ekcs> 1. add to existing HA scenario 00:42:42 <ekcs> 2. add new HA scenario to exsting gate job 00:42:44 <ramineni_> actually im confused, i thought we need both , we need to add tempest tests for HA anyway right , even if we add a new job 00:42:47 <ekcs> 3. add totally new gate job 00:43:28 <ekcs> and I think we’re saying (1) won’t work. 00:44:10 <ramineni_> i mean add a new job to test HA , where we test all HA scenarios 00:45:07 <ramineni_> for multinode 00:46:00 <ekcs> ramineni_: got it. what’s the benefit of adding new gate job over simply adding scenarios to existing gate job? so that we can get the right multinode setup? 00:46:34 <ramineni_> ya, to run on multinode 00:47:28 <ekcs> ok. I’m just thinking might it be possible to setup multinode setup within the scenario similar to the way we currently do it by starting a replica. 00:47:50 <ekcs> so 1. is that possible, and 2. is that a good idea or not. 00:48:33 <masahito> I think 2 is better to manage multinodes. 00:49:23 <masahito> we can launch multi PE nodes in creating devstack phase. 00:49:57 <ramineni_> masahito: but is it possible to start on different nodes in devstack script 00:50:08 <ekcs> right, that’s the main difference: launch multiple nodes in the scenario, or launch multiple nodes in devstacking. 00:50:48 <masahito> ramineni_: gate only accepts one *VM* for devstack but we have dse_id flag. 00:51:09 <ramineni_> oohok, like we do in HA now right 00:51:19 <masahito> yes. 00:51:27 <ramineni_> ok 00:52:20 <masahito> of course, in HA tests like failover, we need to manage the process in tempests. 00:52:20 <ekcs> ok why don’t we think more on this and maybe decide next week whether we want to add scenarios to current gate job or add a whole new gate job? 00:52:47 <masahito> ekcs: right. 00:53:12 <ekcs> aimeeu how are you feeling about ramping up on Congress so far? Anything you're blocking on or having trouble finding help with? 00:53:17 <ramineni_> adding HA tempest tests in assigned to aimee right 00:53:55 <aimeeu> ekcs: learning a lot but just frustrated at own lack of progress. 00:54:55 <aimeeu> ramineni: I unassigned myself from the basic tempest tests - not ready for that yet and can't get tempest working locally 00:55:53 <ekcs> aimeeu: ok I think you’re ramping up great! do ping me or others if you’re blocked on something. 00:56:11 <ramineni_> aimeeu: ok 00:56:36 <ekcs> 4 minutes. anything else or should we end meeting? 00:56:42 <ramineni_> aimeeu: you can ping me in case you want some help in getting tempest working 00:57:03 <aimeeu> ekcs: thanks! probably the best approach for me would be writing unit tests or implementing a new, simple piece of functionality 00:58:01 <ekcs> aimeeu: are you looking for the next suitable thing to take on? 00:58:51 <aimeeu> I have some low-hanging fruit but I was thinking it may be easier to work on a new easy piece of functionality rather than fixing something 00:59:32 <ekcs> aimeeu: got it that makes sense. let’s talk about it offline then since we’re out of time here. 00:59:36 <aimeeu> like adding and testing a method... but I don't want to take time away from what needs to go in Newton-3 00:59:44 <aimeeu> ekcs: perfect 00:59:59 <ekcs> all right thanks everyone! 01:00:04 <ekcs> have a great week! 01:00:06 <ekcs> #endmeeting