00:02:47 #startmeeting congressteammeeting 00:02:48 Meeting started Thu Nov 3 00:02:47 2016 UTC and is due to finish in 60 minutes. The chair is ekcs. Information about MeetBot at http://wiki.debian.org/MeetBot. 00:02:49 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 00:02:51 The meeting name has been set to 'congressteammeeting' 00:04:13 Hi all - on mobile and here to keep track of what's going on 00:04:35 hi aimeeu! 00:04:50 hi all 00:04:57 I'm still recovering from the Summit! 00:05:19 so it seems we have very small group today. masahito told me he’s also on mobile and don’t expect to participate much. I don’t see ramineni guessing she’s recovering from the traveling. 00:05:42 but let’s dive in. 00:06:16 topics I have are: 00:06:17 1. summit wrap-up 00:06:18 2. robustness improvements and bug fixes 00:06:18 3. other ocata priorities 00:06:21 anything else? 00:06:34 LGTM 00:07:13 #topic summit wrap-up 00:08:37 I’ll be sending around a recap over ML. but we had a very productive summit with lots of feedback and use case directions from the summit. will probably have stuff to work on for many cycles to come. 00:09:16 Yes I agree. Good sessions. 00:09:37 +1 00:09:40 pretty nice publicity with the keynote demo. and came away with some priorities to focus on in ocata. 00:09:49 any other comments? 00:10:50 It was great having Nate, Adam, and Bryan talk about such concise use cases. That was a huge help! 00:11:09 +1 00:11:43 +1 00:11:46 well moving on then. 00:12:07 #topic robustness improvements and bug fixes 00:13:18 for ocata priorities, we can probably break them down between robustness improvements / bug fixes vs new features (others) 00:15:12 I’ve been registering bugs for robustness and bug fixes. 00:16:14 renaming topic. 00:16:24 #topic ocata priorities 00:17:24 So based on our discussions at the summit it seems we want to focus on robustness and bug fixes plus out-of-the-box policies and CIDR arithmetic builtin. 00:18:02 I’ve been registering bugs and setting approximate targets. 00:18:11 https://launchpad.net/congress/+milestone/ocata-1 00:18:14 https://launchpad.net/congress/+milestone/ocata-2 00:18:23 https://launchpad.net/congress/+milestone/ocata-3 00:19:53 the ocata-1 targets may be pretty aggressive because it’s in a couple weeks. 00:21:01 but anyway please take a look at the bugs, grab ones you want to work on, and register additional bugs for robustness and fixes and other ocata priorities. 00:21:39 got it 00:22:40 a couple things I’ll highlight 00:23:43 this is a critical bug that we didn’t quite fix in Newton. #link https://bugs.launchpad.net/congress/+bug/1612008 00:23:43 Launchpad bug 1612008 in congress "HAHT - Update synchronizer for distributed deployment" [Critical,Triaged] - Assigned to Anusha (anusha-iiitm) 00:24:53 I realized just this week it wasn’t completely fixed. So I added a new comment on suggested fix and re-opened it. It should be fairly straightforward. 00:26:41 any thoughts or comments or discussion on ocata priorities? 00:27:04 Nope 00:27:52 Looks good 00:28:34 cool. ok so I don’t have much else to discuss. let’s open it up then. 00:28:40 #topic open discussion 00:29:56 Looking thru open changes, I see a couple from Aug for Gate troubleshooting. Need to keep those? 00:30:00 https://review.openstack.org/#/q/project:openstack/congress+status:open 00:30:56 No I think I can abandon them. 00:31:01 I’ll do that. 00:31:40 we had a discussion about this patch before the summit: https://review.openstack.org/#/c/376166/ 00:32:25 I think thinrichs said we should make this fix if add/delete datasource does not immediately take effect on separate node. 00:33:43 and indeed as it stands right now add/delete ds does not take effect until next periodic sync. 00:34:26 So if alice adds a datasource and then lists the datasource, she doesn't see her new datasource necessarily (even if she's pinned by haproxy to the same Congress instance). 00:34:38 Does that happen even in a single-node deployment then? 00:34:52 but instead of leaving that as is and backporting this patch (or a minimal verison of it) it may make more sense to trigger the sync so that a change takes effect. 00:35:10 thinrichs: correct. does not affect single-node deployment. 00:35:33 Why doesn't it affect single-node deploy? 00:35:56 I’ve registered a bug for it. #link https://bugs.launchpad.net/congress/+bug/1638729 we could decide to backport that fix to newton. 00:35:56 Launchpad bug 1638729 in congress "datasource add/delete not effected until next sync" [High,New] 00:36:52 thinrichs: the code today triggers a sync right away if the datasourecs are on the same node as the API processing the request. 00:36:53 https://github.com/openstack/congress/blob/master/congress/dse2/dse_node.py#L748 00:37:11 ekcs: ok—so it's that we're not telling all the nodes to pick up the change 00:37:25 I think it was one of those things we put off deciding in terms what should happen in multi-node case. 00:37:58 If the user lists the datasources, does she see the new one (always), even if it hasn't been created on a separate node? 00:38:13 i.e. is the list of datasources coming from the DB or from oslo-messaging? 00:39:47 It's coming from db .. 00:39:48 seems list comes from db so it would show up. 00:40:35 but if I create a ds and then ask for it’s status, I’d get 404. 00:41:17 It takes 30 seconds to pull the data, which is really the first time the datasource is valuable. 00:41:40 Obviously it'd be better for everything to work faster, but I wouldn't think this fix needs to go into Newton. 00:41:52 true. 00:42:36 it’s not that on it’s own it has to go into newton. rather, if starting from the premise that we need to get at least one of the two into newton, then the sync fix i think makes more sense. 00:42:57 than this fix: https://review.openstack.org/#/c/376166/ 00:43:14 just my current thoughts. 00:43:52 Not sure I see a need to put either fix into Newton. 00:44:21 Instead of the first data pull taking 30 seconds from the time the datasource gets added, it takes 30 seconds until the first sync and another 30 seconds to do the first pull. 00:44:52 I think the bigger problem is delete. that’s why two weeks ago you thought we need to put the fix in. 00:45:21 If i delete a ds and then add again after 10 seconds, I can’t add because it sees a conflict in service name. 00:48:17 anyway feel free to jump in with other topics and thoughts. 00:48:44 That's not great UX. 00:48:48 ekcs: your call whether a fix needs to go into Newton for it. 00:49:30 thinrichs: yea I’ll think more about it. 00:49:51 Nothing else from me. 00:50:02 anyone else still jetlagged from the summit? 00:50:32 yes - wiped out even with 2 days vacation 00:50:48 i’m having a hard time sleeping in to catch up on sleep haha. 00:50:56 lol 00:52:17 anyway I don’t have anything else we need to talk about right now. I’m sure we’ll have discussions over launchpad and gerrit on specific items. 00:52:30 we can end early with no other topics. 00:53:07 I've slept like a champion since I've been back. Kids will do that to you. :) 00:53:23 See you all later! 00:53:26 lol! 00:53:44 Have a nice evening/morning/afternoon! 00:53:48 good to have everyone back safe and sound. see you all next week then? 00:53:53 aimeeu: yup you too. 00:54:53 #endmeeting