15:00:18 #startmeeting oslo 15:00:20 Meeting started Mon Apr 23 15:00:18 2018 UTC and is due to finish in 60 minutes. The chair is bnemec. Information about MeetBot at http://wiki.debian.org/MeetBot. 15:00:21 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 15:00:24 The meeting name has been set to 'oslo' 15:00:27 courtesy ping for amotoki, amrith, ansmith, bnemec, dansmith, dhellmann, dims 15:00:27 courtesy ping for dougwig, e0ne, electrocucaracha, flaper87, garyk, gcb, haypo 15:00:27 courtesy ping for jd__, johnsom, jungleboyj, kgiusti, kragniz, lhx_, raildo 15:00:27 courtesy ping for redrobot, sileht, spamaps, sreshetnyak, stephenfin, stevemar, therve 15:00:27 courtesy ping for thinrichs, toabctl, zhiyan, zxy, zzzeek 15:00:40 o/ 15:00:42 hi 15:00:43 o/ 15:00:49 hi! 15:00:55 o/ 15:01:02 o/ 15:01:20 o/ 15:02:09 #topic Red flags for/from liaisons 15:02:24 Nothing from Cinder. 15:03:37 I can't think of anything we've done that should cause anyone grief, so we can probably move on then. 15:03:47 #topic Releases for Rocky 15:04:10 The usual set last week. Expect a bunch more this week for the eventlet uncapping changes. 15:04:52 I also need to go through the stable branches and release those. I'm pretty sure there are some changes piled up on them. 15:05:11 #topic Action items from last meeting 15:05:24 "finish merging eventlet uncapping patches" 15:05:29 I believe that's done. 15:05:43 I didn't see any more oslo projects in the topic. 15:05:49 https://review.openstack.org/#/q/status:open++topic:uncap-eventlet 15:05:58 "Oslo cores to single approve python 3 tox changes" 15:06:23 Also done. I +1'd the project-config patch Friday to switch all of our repos over to run gate jobs using python 3. 15:06:33 I didn't notice if it merged yet, but expect that to be coming soon. 15:06:55 "fix up any python 3 tox changes that need additional work" 15:07:02 Must have been done since we merged all the things. :-) 15:07:09 "bnemec to check on missing onboarding session" 15:07:18 Done: https://www.openstack.org/summit/vancouver-2018/summit-schedule/events/21645/oslo-project-onboarding 15:07:32 Apparently there was a problem on the backend that caused it to get lost. 15:07:52 But it's back now so we should be good to go. 15:08:05 "kgiusti to look into pip-check-reqs with pip 10" 15:08:17 https://github.com/r1chardj0n3s/pip-check-reqs/issues/17 15:08:37 Looks to me that this project may be pushing up the daisies 15:09:08 If that's indeed the case, what should we do going forward? 15:09:14 Hmm, yeah. 15:09:27 drop the test, or fork and fix (and own) :( 15:09:54 I guess it depends on how much we value the ability to automatically check for missing reqs. 15:09:55 I did not look for a supported alternative 15:10:46 I would imagine if folks found it valuable there'd be an effort (or an alternative)? 15:11:04 seems like a pretty nice tool to have IMHO 15:11:34 kgiusti: I'm guessing the alternative is pipfiles 15:11:53 (side note: morded's recent comments on same on openstack-dev are worth a read) 15:12:22 stephenfin: ah - didn't see that... 15:13:07 tbh, I don't think anyone is really using it. Only today I saw a patch adding pika to a project because oslo.db (I think?) dropped it 15:13:18 Does that have functionality for checking missing requirements? 15:13:44 oslo.messaging dropped pika because we didn't think anyone was using it. :-/ 15:14:12 bnemec: It works somewhat differently. You've got the handwritten Pipfile to track what you think are your dependencies (usually versioned in a range or unversioned) 15:14:28 fyi: http://lists.openstack.org/pipermail/openstack-dev/2018-April/129187.html 15:14:28 then a Pipfile.lock file which has _all_ the dependencies + specific versions 15:14:51 aye, mordred really details it well in that post 15:15:02 Yeah, I've seen the thread, but I have to admit I haven't gotten all the way through it. 15:15:18 I guess it doesn't solve the problem but it would mitigate it, assuming you're using Pipfile.lock 15:15:27 It's kind of above my pay grade as far as requirements management goes. :-) 15:15:42 and I only care because of pbr :) 15:16:18 pipfiles solve a different problem than this 15:16:25 they solve the problem we've solved by using a constraints list 15:16:35 the problem we have is that we depend on things that we don't realize 15:16:40 pika is a great example 15:16:58 why is anyone using pika if oslo.messaging's driver is no longer supported? 15:17:04 are they using it directly? 15:17:20 That's a good question 15:17:23 or are they using it indirectly via another project (taskflow?) that doesn't have the dependency listed? 15:18:05 or that *does* have it listed but uses an extras entry and they don't know how to use that? 15:18:08 dhellmann: I'm not sure. It was a random networking project 15:18:29 pika appears in the lower constraints for *tons* of projects 15:18:30 I only saw it because I was reviewing a sphinxcontrib-apidoc patch 15:18:40 http://paste.openstack.org/show/719753 15:19:08 dhellmann: http://codesearch.openstack.org/?q=pika%3A%2F%2F&i=nope&files=&repos= 15:19:29 leads me to believe these pika deps are not oslo.messaging related 15:19:51 since there's no obvious use of the pika:// o.m. transport driver 15:20:24 Or they didn't document it 15:20:43 I see it in a bunch of charm files; I wonder if charms are configuring oslo.messaging to use it? 15:20:47 http://codesearch.openstack.org/?q=pika&i=nope&files=requirements.txt&repos= 15:21:15 networking-cisco seems to use it directly 15:21:18 http://codesearch.openstack.org/?q=pika&i=nope&files=.*%5C.py&repos= 15:21:51 dhellmann: Ha, that was the one I reviewed :) 15:21:55 anyway, I didn't mean to side-track things 15:21:55 https://review.openstack.org/557307 15:22:08 yeah, likewise 15:22:12 that's a possibility, but we've never seen activity on the pika driver in terms of dev, bugs, etc beyond the devs IIRC 15:22:39 Ah, good, so they're not using the o.m driver at least. 15:22:52 It looks like charms are using pika directly too. 15:23:12 http://git.openstack.org/cgit/openstack/charm-ceilometer/tree/charmhelpers/contrib/openstack/amulet/utils.py 15:23:13 so maybe the charms use it because that's how juju works? I don't know 15:23:42 "Establish and return a pika amqp connection to the rabbitmq service running on a rmq juju unit." 15:23:45 Sounds like it. 15:24:00 yep 15:24:29 I wonder why so many projects have it in their lower-constraints.txt list now, but that may be because they're using an older version of oslo.messaging that still has it as a dependency 15:25:29 It's possible. Maybe we stopped syncing requirements before we removed pika from o.m. 15:26:22 well, those lower-bounds can be pretty low in some cases 15:27:09 It looks like quite a few of them only have it in their test-requirements too. 15:27:49 looks like no one hardcoded pika as an oslo.messaging transport, but perhaps the deps are there to allow deployers to configure pika via transport_url cfg item? 15:28:40 Oh, actually all of the test-requirements are also charms. 15:29:00 if that's the case, they'd need to modify the config to use rabbit in transport_url instead - as described in the release notes. 15:29:46 Assuming they've been ignoring the big scary log message telling them to use rabbit for the last two releases.... 15:29:49 Okay, well maybe we need to pick this up after the meeting. 15:29:57 +1 15:30:10 #action Investigate why pika is in so many lower-constraints files 15:30:29 Have we settled the missing req issue :) ? 15:30:55 should we at least remove the test from tox.ini? 15:31:12 I'm kind of inclined to leave it be for the moment and see if anyone complains or provides a fix PR. 15:31:26 Technically the issue has only been open a week. 15:31:30 ok, let's do that then. 15:31:42 I'd kill it as it seems like cruft. However, I'm happy to be overruled :) 15:31:46 Although given that Richard seems to have kind of disappeared for the past couple of years I don't feel great about the chances of it getting fixed. 15:32:51 I guess I don't feel strongly about it. Maybe start an ML thread and see if anyone cares enough about it to own it? 15:32:55 If not, remove. 15:35:28 here's the source of those lower-constraints entries: "Collecting pika-pool>=0.1.3 (from oslo.messaging==5.36.0)" 15:35:38 #action Send email about pip-check-reqs to gauge interest in keeping it 15:35:42 so those are OK 15:36:31 Cool 15:37:15 #topic PTG attendance 15:37:32 The Foundation wants to know if we need a room at the PTG in Denver. 15:37:57 I realize it's pretty early, but I assume enough of us will be there that I should request one? 15:38:19 Denver? I thought it was Berlin.... 15:38:32 That's the Forum. 15:38:41 ack - thanks 15:39:05 I have registered the Cinder team. I am assuming we will have enough people again. 15:39:38 Huh, I can't find the email that announced it. 15:39:42 I'll definitely be in Denver in september 15:39:45 week of the 10th I think 15:39:58 Oh, they spelled out PTG: http://lists.openstack.org/pipermail/openstack-dev/2018-April/129564.html 15:40:13 Yeah, the 10th-14th. 15:40:25 yeah, we're also likely to see that rebranded since we'll have some operator sessions going on that week, too 15:40:53 Note the early deadline for cheap-ish tickets too. 15:41:09 After May 11 it costs twice as much. 15:41:22 register early and often 15:41:24 Even the early bird rate is double what it used to be. 15:41:34 yes, that's part of making the event more sustainable 15:41:45 still a work-in-progress 15:41:59 Yeah, I can't imagine how $100 covered everything. 15:42:03 Obviously it didn't. :-) 15:42:10 it didn't, that sort of the point :-) 15:42:11 yeah 15:42:39 Okay, I'll ask for a room. I _assume_ I'll be there too, although I almost didn't go to the last Denver PTG so who knows? 15:42:59 Which is funny because it's geographically the closest to me. 15:43:10 #action bnemec to request room at Denver PTG 15:43:17 I'm sure we can find things for bnemec to do to make it worth the trip 15:44:23 Sounds good 15:44:25 #topic oslo.config driver update 15:44:47 I have forgotten when this meeting is held, but I was wondering what the status on the pluggable drivers is. 15:45:12 I looked at the existing patches a bit last week but they looked pretty preliminary so far. 15:45:12 it's on tuesdays at the same time as this meeting 15:45:22 the work done so far is quite early, yes 15:45:42 I just wanted to follow up because I know we wanted to try to get that in as early in the cycle as possible. 15:45:46 we have a new contributor joining that little team, and I'm hoping for more significant contributions soon 15:45:54 yes, progress has been slow 15:46:22 I obviously haven't been very engaged with it myself. :-/ 15:46:33 https://review.openstack.org/562746 is the start of a driver 15:47:35 I feel like maybe that's something we would want to try to merge by R-2? Is that reasonable and/or possible? 15:47:51 it is a reasonable request (the original goal was r-1) 15:48:13 I'm hesitant to say whether it's likely, since I'm not doing the work myself 15:48:35 Maybe I should just join the meeting tomorrow. 15:48:46 that would be good 15:48:53 Probably not going to nail anything down without everyone present anyway. 15:48:59 right 15:49:10 #action bnemec to start attending oslo.config drivers meeting 15:49:43 dhellmann: Okay, thanks for the update. 15:49:53 #topic Open discussion 15:50:21 That was it for the agenda. Anything else in the last 10 minutes? 15:50:23 I could use another review on https://review.openstack.org/557516 15:50:38 it's not pressing, but it would be nice to clear it off the board 15:50:50 this one, too: https://review.openstack.org/562724 15:51:59 That one was previously approved, so we could probably just go ahead with it. 15:51:59 bnemec: Hi Ben 15:52:06 namnh: Hi 15:52:11 bnemec: All of your comments was solved, and the output was chageds as your expectation [1], could you take a look the patch again [1] https://review.openstack.org/#/c/526314/ 15:53:14 namnh: Okay, will try to take another look this week. 15:53:41 bnemec: cool, thanks 15:55:10 Anything else? 15:55:34 that's it from me 15:55:46 Nothing from me. :-) 15:56:05 Okay, thanks for joining everyone. 15:56:06 #endmeeting