19:00:00 <smcginnis> #startmeeting releaseteam
19:00:01 <openstack> Meeting started Thu Jul 11 19:00:00 2019 UTC and is due to finish in 60 minutes.  The chair is smcginnis. Information about MeetBot at http://wiki.debian.org/MeetBot.
19:00:02 <openstack> Useful Commands: #action #agreed #help #info #idea #link #topic #startvote.
19:00:04 <ttx> ohai
19:00:05 <openstack> The meeting name has been set to 'releaseteam'
19:00:07 <smcginnis> Nailed it!
19:00:13 <smcginnis> ping - ttx dhellmann diablo_rojo hberaud evrardjp armstrong
19:00:21 <armstrong> o/
19:00:32 <smcginnis> #link https://etherpad.openstack.org/p/train-relmgt-tracking Agenda
19:00:44 <fungi> aloha
19:01:28 <smcginnis> OK, I guess we can get going.
19:01:40 <smcginnis> I will be your substitute tonyb for the today.
19:01:50 <ttx> missing the accent tho
19:02:02 <smcginnis> We are around line 227 in the etherpad.
19:02:03 <smcginnis> ;)
19:02:21 <smcginnis> #topic Train membership freeze preparation
19:02:28 <smcginnis> #link https://etherpad.openstack.org/p/train-membership-freeze
19:02:34 <ttx> OK so I ran the analysis in preparation for the freeze next week
19:02:42 <ttx> err, week after
19:02:47 <diablo_rojo> o/
19:02:54 <ttx> #link https://etherpad.openstack.org/p/train-membership-freeze
19:03:08 <ttx> On that page you can see Deliverables declared in governance but missing a train deliverable file
19:03:35 <ttx> For those, their PTLs/release liaisons need to decide whether to make it a part of Train or not
19:03:43 <ttx> (4 actions possible, listed at the top)
19:03:51 <smcginnis> Not directly related, but I wonder if we need to raise concerns about ajutant to teh TC.
19:04:03 <ttx> We can find a bunch of things that were skipped in stein already
19:04:08 <ttx> including Adjutant yes
19:04:24 <ttx> The "new" ones are probably just oversights
19:04:45 <ttx> So the way we do this is generally to engage with PTLs/liaisons on those and ask
19:05:06 <smcginnis> Do we need to be as concerned about the tempest plugins?
19:05:08 <openstackgerrit> Merged openstack/releases master: Release python-keystoneclient 3.20.0  https://review.opendev.org/670187
19:05:15 <ttx> And for those with no solution once we hit milestnoe-2, we just skip in Train
19:05:31 <ttx> smcginnis: tempest plugin we could automatically set to cycle-automatic
19:05:39 <ttx> since that is what we did to the others
19:06:18 <ttx> Helm we can also safely skip, since it would be independent if ever released
19:06:22 <smcginnis> That would make sense to me.
19:06:34 <ttx> It's the other we need to engage with
19:06:58 <ttx> Not sure how to best engage
19:07:04 <ttx> We can split teh list by affinity
19:07:12 <smcginnis> ML? Or should we try to ping some like TheJulia for the Ironic ones?
19:07:14 <ttx> or just dump a bunch of emails on -discuss
19:07:32 <ttx> and EmilienM on TripleO
19:08:00 <ttx> I propose a first round of IRC pings today/tomorrow, followed by emails early next week
19:08:21 <ttx> hmm unless that was material for the next release countdown
19:08:53 <diablo_rojo> Sounds like a solid plan to me.
19:08:54 <smcginnis> I think we've had better responses with more directed posts to openstack-discuss.
19:09:18 <smcginnis> Should we split these up? Or have one volunteer handle them all since it will mostly be cut and paste?
19:09:43 <ttx> actually we planned to mention those in next countdown
19:09:56 <ttx> so maybe informal pings until then
19:10:16 <smcginnis> OK, we can just go with that plan for now. Then if we don't get a response, follow up with directed posts and more pings afterwards.
19:10:33 <smcginnis> We at least have the data, so thanks ttx for putting that together.
19:10:55 <smcginnis> Anything more on this for now then?
19:10:59 <diablo_rojo> I can help ping some people
19:11:18 <diablo_rojo> I'll throw my name down on some in the etherpad
19:11:24 <smcginnis> Great, I think IRC is probably the most likely way to get a response.
19:11:32 <ttx> The other big thing is generating the list of intermediary-released services that haven't done a release yet. Remind teams that intermediary-released services that have not done a release by milestone-2 should be switched to the cycle-with-rc model.
19:11:36 <diablo_rojo> smcginnis, yep :) That was the plan.
19:11:39 <openstackgerrit> Merged openstack/releases master: Release os-brick 2.3.8  https://review.opendev.org/669764
19:11:48 <ttx> My patch has updated instructions for that
19:12:10 <ttx> https://review.opendev.org/#/c/669186/2/doc/source/reference/process.rst
19:12:17 <smcginnis> And IIRC, Tony had at least the start of updates to make that easier to script?
19:12:39 <ttx> dunno, just updated the list command
19:12:41 <smcginnis> Oh, I think I was thinking of yours.
19:12:59 <ttx> the idea being to publish the list for the next countdown email
19:13:16 <ttx> but that data can be generated when the email needs to be sent
19:13:36 <ttx> so that is all
19:13:55 <smcginnis> OK, thanks
19:14:04 <smcginnis> #topic tempest-plugin releases
19:14:21 <smcginnis> I think I may have added this a couple weeks ago but we didn't have a meeting or something.
19:14:28 <smcginnis> We have a couple things going with tempest plugins.
19:15:03 <ttx> yeah, I was wondering a bit too
19:15:05 <smcginnis> There's the work to make them automatic at the end of the cycle with the idea that they are just a flag to indicate what version of the plugin is meant for what version of tempest.
19:15:13 <ttx> Looking at https://review.opendev.org/#/c/670172/ this morning
19:15:23 <smcginnis> But then there are some teams that are doing multiple releases and even stable releases of their plugins.
19:15:35 <smcginnis> So I think we need to decide how we really want to handle those things.
19:15:48 <ttx> I /think/ what we did in the past is allow new releases on the latest branch
19:15:59 <ttx> it's a bit weird because they have series, but no branch
19:16:15 <smcginnis> ttx: With the cycle-automatic changes, there's nothing that would prevent someone from doing their own releases ahead of time, right?
19:16:15 <ttx> so "stein" is their master
19:16:22 <smcginnis> Yeah, good point about no stable branch.
19:16:23 <fungi> the plugins are in their own repos separate from the services they're testing, right?
19:16:29 <smcginnis> Yep
19:16:32 <ttx> smcginnis: no, they would just alter the release model atr the same time they are doing release
19:16:47 <ttx> It's just a "default" model in case they don't
19:16:48 <fungi> i thought the idea was tempest plugin repos would remain branchless like tempest itself?
19:17:05 <ttx> fungi: yes, but they still have "stein" deliverables
19:17:20 <fungi> ahh, i see
19:17:29 <smcginnis> OK, so if we set them to automatic and then they decide they want to do a milestone-2 release, they would just have to switch over to cycle-with-intermediary - right?
19:17:39 <ttx> yes
19:17:49 <ttx> fungi: so you can't do a new... rocky release say
19:18:01 <ttx> but at this point you can either do another stein release, or the first train release
19:18:12 <fungi> i guess there's no explicit metadata right now to indicate intentionally branchless repos?
19:18:19 <fungi> or is there?
19:18:31 <ttx> nothing except the fact that they have no stable branch in an "old" series
19:18:47 <ttx> It's always been a bit of a one-off
19:19:03 <smcginnis> When the idea of even tagging tempest plugins was raised on the ML, it was very much opposed by some.
19:19:07 <ttx> But every time I dive into it, the can of worms ends up eating me alive
19:19:23 <ttx> and I just defer to someone that knows better
19:19:59 <ttx> So yeah, the reality is, they are more of a single channel, with a pointer having the name of the series
19:20:01 <smcginnis> Theoretically, and very of the plugin should work against any version of a deployment. It was more to make sure the plugins were compatible with the release of tempest itself being used.
19:20:07 <ttx> instead of muletiple stable channels
19:20:26 <smcginnis> s/and very/any version/
19:20:38 <smcginnis> Typing not working well for me today.
19:20:39 <ttx> We could achieve the same with _independent + a documentation saying which version is to be used with which series
19:21:22 <smcginnis> That actually seems a little simpler. Though we lose some of the implicit documenation of seeing with cycle the release was part of by which deliverable it was in.
19:21:36 <smcginnis> And it showing up in https://releases.openstack.org/stein/index.html#tempest-plugins
19:21:41 <ttx> also I think tempest ends up magically using the right thing
19:21:57 <ttx> so yeah... probably better to leave it alone
19:22:08 <smcginnis> Just what I was thinking.
19:22:18 <smcginnis> Let teams do extra releases if they feel a need for them.
19:22:32 <smcginnis> Switch the others to automatic for the ones that don't want to care or worry about it.
19:22:34 <ttx> but we should find a way to document this, because it confuses the hell out of me every time
19:22:37 <fungi> right, if i recall the discussion (from... denver ptg ii?) it was that folks are runnnig older tempest releases for refstack/interop conformance and need to know what plugin versions to use with them
19:22:53 <smcginnis> Yeah, something like that.
19:23:10 <ttx> So we should approve https://review.opendev.org/#/c/670172/4
19:23:29 <smcginnis> I guess so.
19:23:30 <ttx> despite its confsing commit message, it's actually right
19:23:53 * ttx +2s
19:24:09 <smcginnis> So it will show up under stein at least, just not on a stable/stein branch.
19:24:17 <smcginnis> Which sounds to be their intent.
19:24:27 <smcginnis> And then they will do a release later for train.
19:24:30 <ttx> yes. And we should definitely refuse a bump on an old series
19:24:49 <smcginnis> And if they find something else they think they need in stein, that's just not going to be possible.
19:24:51 <ttx> once you start callign things train, you can't go back
19:25:06 <smcginnis> One way train.
19:25:15 <ttx> slow clap
19:25:20 <smcginnis> :P
19:25:39 <fungi> this is why i was thinking if projects could be flagged that they're explicitly branchless, it gives you an opportunity to reject them via ci jobs when they try to bump older series (or accidentally add a branch parameter for that matter)
19:26:01 <ttx> https://giphy.com/gifs/mlb-tips-jonathan-villar-ATlL0sKCXwsWk/fullscreen
19:26:08 <smcginnis> We do reject branch requests on those.
19:26:26 <fungi> but if it doesn't come up often, then maybe not worth the effort to automate
19:26:31 <smcginnis> Haha, the gif game has made its way here too.
19:26:48 <ttx> can't put the gif game back into the bottle
19:27:02 <smcginnis> :)
19:27:21 <smcginnis> OK, so there's still confusion, but at least less confusion than before.
19:27:24 <smcginnis> Good enough for now.
19:27:43 <smcginnis> Those were the only two things on the agenda. Any other topics?
19:28:01 <ttx> I'll likely miss next two meetings, so if you have questions on PROCESS tasks...
19:28:07 <ttx> https://github.com/openstack/releases/blob/master/doc/source/reference/process.rst#between-milestone-1-and-milestone-2
19:28:11 <ttx> ew
19:28:25 <smcginnis> I am out the next two weeks as well.
19:28:30 <ttx> damnit google
19:28:36 <smcginnis> Should be able to check in from time to time.
19:29:22 <ttx> https://opendev.org/openstack/releases/src/branch/master/doc/source/reference/process.rst#between-milestone-1-and-milestone-2
19:29:23 <ttx> there
19:29:54 <ttx> got to prefer funky syntax highlight
19:30:17 <smcginnis> Is Tony back next week?
19:30:39 <ttx> yes
19:30:45 <smcginnis> Most of those tasks are related to the countdown email, so as long as that gets sent.
19:30:53 <ttx> All tasks listed affect the email he needs to send
19:31:05 <ttx> indeed
19:31:40 <ttx> It's an important one, since we need to warn people of what we'll do if they don;t act
19:31:56 <diablo_rojo> So should I just ping all the deliverables with missing deliverable file folks? Or does someone want half? :)
19:32:02 <ttx> We missed a couple warnings last milestones and that delayed the whole thing
19:32:04 <smcginnis> Maybe we should all set reminders to remind Tony next week.
19:32:39 <ttx> diablo_rojo: I would just ping them on release channel, pointing them to the etherpad
19:32:45 <ttx> If they are not here, I would not chase them down
19:32:55 <fungi> i plan to just turn my entire to do list into reminders for tony next week ;)
19:32:59 <ttx> They can take the next train
19:33:23 <diablo_rojo> ttx, noted
19:33:25 <diablo_rojo> can do
19:33:26 <fungi> the train with no name
19:33:49 * fungi is, like neil young, riding through the desert on a train with no name
19:33:54 <ttx> https://tenor.com/view/train-nowhere-gif-7641604
19:34:10 <diablo_rojo> http://gph.is/22ZevDy
19:34:30 <diablo_rojo> read as 'no deliverable file'
19:34:52 <fungi> oh, hah, that song wasn't even neil young. certainly sounded like him though
19:35:02 <smcginnis> Hopefully not https://tenor.com/view/tight-tightfit-train-gif-4902750
19:35:44 <smcginnis> OK, anything else?
19:35:54 <ttx> nope
19:35:59 <diablo_rojo> None from me
19:36:12 <smcginnis> OK, thanks everyone.
19:36:21 <fungi> my favorite train line terminus is the northbound blue in boston to "wonderland"
19:36:25 <smcginnis> #endmeeting