15:00:36 <smcginnis> #startmeeting releaseteam
15:00:37 <openstack> Meeting started Fri Sep 28 15:00:36 2018 UTC and is due to finish in 60 minutes.  The chair is smcginnis. Information about MeetBot at http://wiki.debian.org/MeetBot.
15:00:38 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
15:00:40 <openstack> The meeting name has been set to 'releaseteam'
15:01:12 <smcginnis> ttx, dhellmann, fungi, armstrong, others.
15:01:19 <smcginnis> Whew, didn't get spam blocked on that.
15:01:20 <ttx> o/
15:01:20 <fungi> heyhey!
15:01:23 <armstrong> Hello
15:01:51 <smcginnis> #link https://etherpad.openstack.org/p/stein-relmgt-tracking
15:01:54 <smcginnis> Around line 42.
15:02:04 <dhellmann> o/
15:02:11 <smcginnis> #topic Release failures
15:02:30 <smcginnis> Mostly since I was out most of the week, I was hoping we could have a quick recap of the release failures.
15:02:43 <smcginnis> And just a quick sanity check that everything has been fixed up.
15:02:45 * dhellmann clears his throat nervously
15:02:54 <smcginnis> dhellmann: It was some package update?
15:03:10 <fungi> twine
15:03:11 <dhellmann> twine version 1.12.0 reported a failure even when it succeeded
15:03:12 <fungi> 1.12.1
15:03:16 <fungi> had the fix
15:03:30 <lbragstad> o/
15:03:31 <smcginnis> Well that doesn't sound like a good thing. :)
15:04:03 <dhellmann> after spending a bunch of time figuring that out, I decided to wait for the new twine release and re-tag all of the releases I had already approved
15:04:09 <smcginnis> dhellmann: I saw you had the tracking etherpad for failures in the tracking etherpad. Anything we need to check in there?
15:04:25 <dhellmann> because there were a lot of them, and some of the artifacts had been uploaded, so rerunning the existing tags would be a lot of work
15:04:40 <smcginnis> Yeah, good call.
15:04:47 <dhellmann> when twine 1.12.1 was released I tested it with the release-test repo, and it worked fine
15:05:04 <dhellmann> so I prepared a single patch to retag all of the deliverables we missed with new patch-level updates to their version  numbers
15:05:09 <dhellmann> those all went through OK
15:05:22 <smcginnis> Thanks for handling all of that.
15:05:24 <dhellmann> I think I missed rally in that set, so we came back later and did a new release of rally separately
15:05:36 <dhellmann> that release actually included a fix or two, so it wasn't a direct retag
15:05:58 <dhellmann> and I have now learned to test the release machinery before approving a bunch of things on a monday morning :-)
15:06:14 <smcginnis> :)
15:07:24 <smcginnis> Just running a quick missing-releases report, there are some older rally ones missing, but otherwise everything looks fine.
15:07:39 <smcginnis> And actually, just the one expected missing and a bunch of older asc files.
15:08:16 <dhellmann> yeah, so we could comment out the missing one or just leave it
15:08:48 <smcginnis> So probably nothing we need to change or document with this whole situation other than maybe be cautious Monday mornings. :)
15:09:25 <smcginnis> #topic PTG proposed changes
15:09:39 <ttx> that's vague
15:09:43 <smcginnis> I sent out the first part of the plan we discussed in Denver for dropping milestone releases.
15:09:48 <ttx> ah, THOSE
15:09:59 <smcginnis> Surprisingly less of a reaction than I expected.
15:10:21 <ttx> mine at least did trigger one answer
15:10:36 <smcginnis> So next step is probably to make the cycle-with-rc change.
15:10:57 <smcginnis> Should we rename the existing model or introduce it as a new one? Probably just rename?
15:11:00 <fungi> i expect the forced tagging of cycle-with-intermediary will be the controversial one, of any
15:11:10 <smcginnis> Yeah, that one will be a bigger one.
15:11:25 <dhellmann> smcginnis : let's make a new one. it will mean fewer updates to existing deliverable files
15:11:48 <dhellmann> we can switch just the stein deliverables and set up validation to disallow the old name in new releases
15:11:54 <dhellmann> or new series I should say
15:11:55 <ttx> rename yes
15:12:11 <smcginnis> I was thinking new one would be safer.
15:12:31 <ttx> ah hm
15:13:28 <ttx> your call
15:13:32 <dhellmann> backwards compatibility ftw?
15:14:35 <smcginnis> OK, how about we start with introducing a new model, and if we find we want to later, we can convert exising -milestone ones over to -rc and get rid of it after the fact?
15:15:07 <dhellmann> well, we're going to want to convert the stein one over, and the older ones don't matter at this point
15:15:38 <ttx> dhellmann: so the trick with a new one is that we'll have to lave old doc around ?
15:15:39 <dhellmann> I just meant we would need fewer exceptions for checking the release model in those older files if we keep the old name
15:15:51 <smcginnis> Yeah
15:15:53 <ttx> leave*
15:15:57 <dhellmann> let me look at the validation quickly, hang on
15:16:27 <dhellmann> ok, the validation relies on the schema.yaml file
15:16:41 <dhellmann> so we can leave the old name in there and remove the docs for it
15:16:54 <smcginnis> Sounds reasonable to me.
15:16:56 <ttx> ok
15:17:07 <dhellmann> then we can add a validation rule that says cycle-with-milestones can't be used in the current series
15:18:07 <dhellmann> we can do that in the validate_new_releases() function
15:18:10 <dhellmann> that's what that's for
15:18:36 <smcginnis> Looks like enough of a plan.
15:18:57 <dhellmann> there are probably some tests to update, too, but it ought to be doable with 1 patch
15:19:06 <smcginnis> Then next step - I can send out the cycle-with-intermediary post early next week.
15:19:16 <smcginnis> That will probably result in more discussion.
15:19:41 <smcginnis> #link
15:19:44 <smcginnis> Bah
15:19:49 <smcginnis> #link https://etherpad.openstack.org/p/stein-relmgt-auto-release-change
15:19:55 <smcginnis> Those drafts again ^
15:21:10 <dhellmann> yes, I think that's the next step
15:21:54 <ttx> I think email 2 can be sent next week
15:22:11 <ttx> maybe also email 3
15:22:26 <smcginnis> I'll shoot for Tuesday or Wednesday.
15:22:53 <armstrong> @smcginnis: So cycle-with-milestones will no longer be use at all
15:23:16 <armstrong> I missed the discussion at Denver
15:23:31 <smcginnis> armstrong: Correct. We're going to try dropping milestone releases for service projects and only require the RC and final releases.
15:23:39 <dhellmann> armstrong : right, cycle-with-rc is a variation on that one
15:23:52 <smcginnis> They can still do earlier beta releases if they have a need, but it isn't going to be something that we are going to enforce or necessarily encourage.
15:23:57 <ttx> smcginnis: we are dropping requirements to tag at milestones, so the name did not make that much sense anymore
15:24:13 <ttx> err armstrong ^
15:24:17 <smcginnis> :)
15:24:37 <armstrong> Ok.  Thanks
15:24:51 <smcginnis> #topic Open discussion
15:25:00 <smcginnis> Anything else we should cover at this point?
15:25:32 <fungi> armstrong: http://lists.openstack.org/pipermail/openstack-dev/2018-September/135088.html was the announcement
15:25:46 <ttx> I don't have anything. Making good progress on other work lines
15:25:53 <dhellmann> since it's long, I wanted to make sure everyone noticed my planned 2 week vacation coming up in R-25 and R-26
15:26:10 <dhellmann> sorry, 25 and 24
15:26:12 <lbragstad> nice - enjoy the time off dhellmann :)
15:26:16 <armstrong> @fungi: great! I will go through it
15:26:20 <ttx> been engaging with teams that publish artifacts in external platforms to expose taht in the project navigator
15:26:20 <dhellmann> lbragstad : thanks!
15:27:09 <smcginnis> Wow, two whole weeks!
15:27:10 <ttx> I'll be heavily traveling (read: in China) over weeks -25 and -26
15:27:34 <smcginnis> I think I have all my travel on the etherpad.
15:27:36 <ttx> (and in Europe)
15:27:54 <smcginnis> But that reminds me, I will be on a plane next Friday at this time.
15:28:08 <smcginnis> Does someone want to run that meeting, or should we just cancel?
15:29:02 <dhellmann> do we have anything critical on the schedule?
15:29:13 <smcginnis> Not really at this point.
15:29:32 <smcginnis> We're still a few weeks out to milestone 1.
15:29:38 <ttx> yay long cycles
15:29:51 <smcginnis> We have a few tasks in progress, but nothing I think is critical for timing right now.
15:30:06 <dhellmann> yeah, so we could just plan on office hours or one of us could run the meeting
15:30:22 <ttx> dhellmann: a way to have perfect 6-month cycles would be to adopt something like https://xkcd.com/2050/
15:30:33 <smcginnis> Let's plan on office hours unless something comes up next week that warrants an actual meeting.
15:31:02 <smcginnis> Hah
15:31:16 <dhellmann> ttx: sounds like UTC to me, frankly
15:31:54 * fungi still remembers swatch beat time
15:32:09 <dhellmann> heh
15:32:12 <smcginnis> OK, we can make a call next week on a meeting. That's all I had for this week's.
15:32:13 <ttx> Summits should be on April 15 and October 15, and teh year and time shoudl organize around them to match when they actually occur
15:32:39 <dhellmann> I keep saying we'd have less trouble coordinating with venues if we just bought one.
15:32:57 <dhellmann> we could rent it out to other foundations to cover the mortgage
15:33:06 <dhellmann> and then they would have to schedule around us
15:33:09 <fungi> the stackdrome
15:33:21 <smcginnis> Are buildings in Detroit still cheap? :)
15:33:34 <fungi> depends on if you need them to have walls and roofs
15:33:39 <dhellmann> I saw a ton of buildings for sale in Glasgow a couple of months ago
15:33:57 <smcginnis> fungi: "Open" source?
15:34:02 <fungi> open air
15:34:22 <fungi> the fifth open
15:34:28 <smcginnis> OK, I think this meeting was done awhile ago now. :)
15:34:30 <smcginnis> Thanks everyone.
15:34:36 <fungi> thanks smcginnis!
15:34:40 <lbragstad> o/
15:34:43 <smcginnis> #endmeeting