00:00:21 <thinrichs> #startmeeting CongressTeamMeeting 00:00:22 <openstack> Meeting started Thu Nov 12 00:00:21 2015 UTC and is due to finish in 60 minutes. The chair is thinrichs. Information about MeetBot at http://wiki.debian.org/MeetBot. 00:00:24 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 00:00:27 <openstack> The meeting name has been set to 'congressteammeeting' 00:01:12 <thinrichs> Did the U.S. people remember the time change? 00:01:27 <thinrichs> Who is here? 00:01:37 <ekcs> I'm here. 00:01:47 <ramineni> hi /o 00:01:53 <pballand> hi 00:01:59 <masahito> I'm here. 00:02:50 <thinrichs> We have a collection of odds and ends for the agenda this week. 00:03:00 <thinrichs> Anyone have anything in particular to discuss? 00:03:35 <thinrichs> #topic Tokyo summit 00:03:48 <thinrichs> Sorry I missed the meeting last week. 00:03:53 <thinrichs> Forgot about the time change. 00:04:07 <thinrichs> I did go ahead and write up a recap of what happened at the Summit and sent it to the ML. 00:04:10 <thinrichs> #link http://lists.openstack.org/pipermail/openstack-dev/2015-November/078578.html 00:04:23 <thinrichs> We have a couple of follow up items that I'll get to later. 00:04:44 <thinrichs> Overall, we had a great time and had many productive discussions. 00:05:05 <thinrichs> We also had our first Congress summit dinner. Thanks to masahito for organizing! 00:05:33 <thinrichs> Any comments/questions about the summit? 00:06:18 <thinrichs> #topic devstack plugin 00:06:38 <thinrichs> The way we run tempest tests using devstack in the gate will no longer be functional in 1 month (Mitaka-1). 00:07:09 <thinrichs> Here's a patch that *should* migrate us over to using the devstack plugin so that our gating continues working properly. 00:07:17 <thinrichs> #link https://review.openstack.org/#/c/244224/ 00:07:38 <thinrichs> If 1-2 people could review it, that would be helpful. 00:08:11 <thinrichs> We'll need to wait on the infra team anyway, but it'd be good to make sure it's right so they can just +2 it. 00:08:18 <thinrichs> I did test it manually, and it seemed to work. 00:08:23 <ramineni> thinrichs: there is one mroe patch for the same https://review.openstack.org/#/c/241969/ 00:09:00 <ramineni> i will abandon it , but dont we need DEVSTACK_GATE_TEMPEST_ALL_PLUGINS for tempest to run plugins 00:09:01 <ramineni> ? 00:09:33 <thinrichs> ramineni: thanks for the pointer. I missed that one. 00:09:45 <thinrichs> I don't know about DEVSTACK_GATE_TEMPEST_ALL_PLUGINS. 00:10:08 <thinrichs> I think we don't want to set DEVSTACK_GATE_TEMPEST_REGEX. 00:10:15 <ramineni> thinrichs: it was a recent change to enable plugins 00:10:28 <ramineni> thinrichs: do we want to run all tempest tests in our gate/ 00:10:30 <ramineni> ? 00:10:30 <thinrichs> When I tried that to disable seemingly irrelevant tests, I got pushback from infra. 00:10:45 <ramineni> thinrichs: oh 00:11:20 <thinrichs> ramineni: so we could try to run just our tests, but it makes some sense to run basic tests to ensure we're not breaking other projects. 00:11:38 <gmann> ramineni: thinrichs that will be needed if congress plugin tests you want to run on gate 00:11:39 <ramineni> thinrichs: ok 00:12:23 <thinrichs> Let's do this: could people look at both patches and leave comments about what we think the right thing to do is? 00:12:50 <thinrichs> I find that devstack and devstack gate aren't well-documented and end up reading through the code to figure out what each variable does. 00:13:04 <ramineni> thinrichs: ya :( 00:13:49 <thinrichs> ramineni: I'm happy to abandon mine too. It would actually be better if I weren't on the critical path for this. 00:14:04 <thinrichs> So let's just wait to see what the comments are and pick whichever one gets this done quicker. 00:14:29 <ramineni> thinrichs: sure, that would be good 00:14:41 <thinrichs> ramineni: thanks for working on this by the way—it's getting to be urgent. 00:15:21 <thinrichs> #topic distributed architecture 00:15:40 <thinrichs> Here are the blueprints we're trying to get covered for the new architecture 00:15:42 <thinrichs> #link https://blueprints.launchpad.net/congress 00:17:02 <thinrichs> I know some of those are completed, but I can't tell which ones. 00:17:10 <thinrichs> Most of the API are done or are close. 00:17:19 <thinrichs> pballand: want to give an update on the DseNode? 00:18:11 <pballand> still in the same spot :( getting it wot work with the fake driver in addition the amqp driver 00:18:28 <pballand> will be working with masahito on it 00:19:24 <thinrichs> pballand: any idea about when it'll be ready? 00:20:06 <pballand> I shouldn’t have any more distractions the rest of the week, so there’s hope by next meeting 00:20:23 <thinrichs> pballand: sounds good. Thanks. 00:20:43 <thinrichs> masahito: I see that the rpcify-table blueprint has a commit that 'partially-implements' the blueprint. 00:21:00 <masahito> yes. 00:21:03 <thinrichs> masahito: what is missing from that change that still needs to be done? 00:21:31 <thinrichs> masahito: in other words, can we mark that blueprint as Implemented? 00:21:58 <masahito> The API model only uses fake 'rpc' method so I use partially-implements. 00:22:47 <masahito> If it's not problem using Implemented, I don't mind use Implemented. 00:23:49 <thinrichs> masahito: understood. I think we've been ignoring the migration of fake rpc to real rpc. Whenever we hook everything together, we'll need to make one change that touches everything. 00:24:08 <thinrichs> masahito: so we could mark that blueprint as Implemented, or leave it as is. Your decision. 00:24:46 <masahito> thinrichs: got it. change to Implemented. 00:24:50 <thinrichs> The only non-cleanup piece I think we're missing is this one... 00:24:51 <thinrichs> #link https://blueprints.launchpad.net/congress/+spec/dist-datasources-on-bus 00:26:05 <thinrichs> I think we can get everything working in unit tests without it, but once we start deploying on multiple processes it may be important. 00:26:11 <thinrichs> Anyone want to volunteer? 00:26:43 <ramineni> thinrichs: i can take it up 00:27:13 <thinrichs> ramineni: thanks! 00:27:21 <thinrichs> ramineni: Let us know if the problem and approach is (not) clear. 00:27:29 <ramineni> thinrichs: but might need some help, so will ping on irc for any issue :) 00:27:38 <thinrichs> ramineni: good plan. 00:27:47 <ramineni> thinrichs: thanks 00:27:47 <thinrichs> ramineni: we're always happy to help. 00:28:04 <thinrichs> Anything else on the distributed architecture? 00:29:18 <thinrichs> #topic python3 00:29:43 <thinrichs> ekcs: how is antlr going? 00:30:41 <ekcs> I think the python3 antlr pieces are ready. but real testing can only be done when datalog and other pieces have been patched to python3. working on those now while antlr3 piece under reivew. 00:30:59 <ekcs> https://review.openstack.org/#/c/242305/ 00:31:30 <ekcs> comments on how the cross-compatibility between python 2 and 3 appreciated. 00:31:53 <ekcs> it's done differently here than in most other cases because we're using third party code that is not cross-compatible. 00:32:31 <ekcs> correction: comments on how cross compatibility of handled here in this patch appreciated. 00:33:10 <thinrichs> I took a look and thought it seemed okay but don't know what best practices for this kind of thing are. 00:33:29 <thinrichs> One thing to keep in mind is that all of OpenStack is saying that we're not supporting python2 any more. 00:33:31 <ekcs> i'm in same boat. 00:33:43 <thinrichs> So we may not care about being compatible with python2 at all. 00:34:26 <ekcs> hmmm. 00:34:30 <thinrichs> The python2 tests are deprecated right now. 00:34:37 <thinrichs> BTW this all is coming from the Python community. 00:34:52 <thinrichs> Python2 has been end-of-lifed for years. 00:35:13 <ekcs> so how do you see us managing the transition in congress? 00:35:17 <thinrichs> It seems they are done supporting it at all, even with security fixes. 00:35:26 <ekcs> like one day just flip py27 to non-voting and py34 to voting? 00:35:32 <thinrichs> ekcs: get python3 tests working, then turn off py2 tests 00:35:34 <thinrichs> ekcs: yep 00:36:36 <ekcs> ok that makes sense. But still since the migration is done piece-meal, we need to commit python3 ready code that in the mean time still pass py27... 00:36:48 <thinrichs> I'd say that makes sense. 00:36:50 <ekcs> but maybe how that's achieved isn't so important because soon we'll ditch py2 00:37:03 <thinrichs> ekcs: probably right 00:37:11 <ekcs> ok 00:37:42 <thinrichs> Other thoughts on py3? 00:38:02 <thinrichs> #topic tempest tests 00:38:21 <thinrichs> It turns out that we haven't been running ANY tempest tests in the gate for a while. 00:38:41 <thinrichs> We've been checking that our changes don't break devstack, but the tempest tests aren't running. 00:38:53 <thinrichs> I think it's been since we moved to using the tempest plugin. 00:39:15 <thinrichs> This isn't such a big deal since we only had 2 really simple tempest tests functioning. 00:39:49 <thinrichs> The weird thing is that when we have devstack installed, the following DOES run the tests... 00:39:59 <thinrichs> $ run_tempest.sh -N congress 00:40:20 <thinrichs> But the way tempest tests are run in the gate, none of our Congress tests actually run.... 00:40:27 <thinrichs> $ run_tempest.sh congress 00:40:51 <thinrichs> The difference is that the latter first creates a virtual environment; the former does not (the -N flag). 00:41:07 <thinrichs> I reached out to Zhenzan, who wrote the tempest plugin earlier this morning but haven't heard back. 00:41:35 <thinrichs> If anyone has some cycles to look at the problem, that'd be great. 00:42:27 <thinrichs> Independently but on the same topic, we should ramp up our tempest tests again to have some good system tests. 00:42:33 <ramineni> thinrichs: do you mean to say , run_tempest.sh congress doesn work locally also? 00:42:45 <thinrichs> Fixing the ones that are flakey, for example. 00:42:57 <ramineni> and un_tempest.sh -N congress works? 00:42:59 <thinrichs> ramineni: That's what I saw earlier today. 00:43:08 <thinrichs> What does un_tempest.sh do? 00:43:31 <ramineni> thinrichs: sorry, typo run_tempest** 00:43:50 <thinrichs> ramineni: got it. Then yes that's what I'm saying. 00:44:10 <ramineni> thinrichs: thats strange .. 00:44:29 <thinrichs> It's happened before. 00:44:38 <ramineni> thinrichs: oh 00:45:25 <thinrichs> The tests would run but then fail b/c we were assuming the presence of a library that was present in devstack but not installed in the virtualenv. 00:45:49 <thinrichs> Maybe something similar this time, to do with the fact that we're using a plugin and tempest needs to find our test suite? 00:45:59 <ramineni> thinrichs: ok , 00:46:17 <thinrichs> ramineni: does this have something to do with that devstack flag for running plugin tests? 00:47:18 <ramineni> thinrichs: im not sure , i have seen it here https://github.com/openstack-infra/devstack-gate/blob/master/devstack-vm-gate.sh#L745 00:47:29 <ramineni> thinrichs: recently added 00:47:55 <thinrichs> I was thinking about DEVSTACK_GATE_TEMPEST_ALL_PLUGINS 00:47:58 <ramineni> thinrichs: but yes, we can give it a try , it says to enable plugins 00:48:34 <ramineni> thinrichs: yes, with plugins the way we run tox is different 00:48:44 <ramineni> it should be tox eall-plugin 00:48:59 <thinrichs> Maybe the thing to do is look at the tox option -eall-plugin target in tempest 00:49:20 <ramineni> thinrichs: otheriwse it runs tox -eall , which doesnt work for plugins 00:49:30 <ramineni> https://github.com/openstack-infra/devstack-gate/blob/master/devstack-vm-gate.sh#L753 00:49:47 <ramineni> thinrichs: according to code here , ALL_PLUGINS should take care of that 00:49:48 <thinrichs> ramineni: maybe that's the solution then and your devstack patch will be doing the right thing. 00:51:00 <ramineni> thinrichs: ya , thats what i heard from qa guys , thats the change required to run plugins 00:51:11 <ramineni> thinrichs: but , ya we can remove the regex and see 00:51:20 <ramineni> to run all tempest tests 00:51:55 <thinrichs> ramineni: Great! Figuring that stuff out takes a bunch of time. 00:52:17 <thinrichs> #topic 00:52:21 <thinrichs> #topic open discussion 00:52:40 <thinrichs> Anyone have anything they want to discuss with the group? 00:54:48 <thinrichs> The Monasca team is putting together a datasource for Monasca. 00:54:49 <thinrichs> #link https://review.openstack.org/#/c/241826/ 00:55:12 <thinrichs> It's great to see them following up on what we discussed at the summit. 00:56:56 <thinrichs> Since there's nothing else, we'll end the meeting. 00:57:02 <thinrichs> Thanks all! 00:57:17 <masahito> bye 00:57:39 <thinrichs> #endmeeting