13:31:31 #startmeeting rpm_packaging 13:31:32 Meeting started Thu Oct 1 13:31:31 2020 UTC and is due to finish in 60 minutes. The chair is jpena. Information about MeetBot at http://wiki.debian.org/MeetBot. 13:31:33 Useful Commands: #action #agreed #help #info #idea #link #topic #startvote. 13:31:35 The meeting name has been set to 'rpm_packaging' 13:31:36 ping toabctl, dirk, apevec, jpena, number80, kaslcrof, rha, hberaud, sboyron 13:31:40 #topic roll call 13:31:52 Remember to add any last-minute topic to the agenda at https://etherpad.opendev.org/p/openstack-rpm-packaging 13:32:08 o/ 13:32:17 #chair dirk 13:32:19 Current chairs: dirk jpena 13:32:41 o/ 13:33:38 #chair hberaud 13:33:39 Current chairs: dirk hberaud jpena 13:35:11 let's start with the agenda 13:35:17 #topic branching? 13:35:41 next week we will have the final RCs, so it might be a good moment for branching 13:35:49 +1 13:36:33 If I'm right only trailing projects will remains branchless for now 13:37:05 and fully brancless projects too 13:37:56 well, we're not really a trailing project anymore? ;) 13:38:20 we could be branching right now I think 13:38:28 we're a SIG now, although there's still one missing review... 13:38:30 alternatively toabctl is currently trying to get testing going 13:38:46 jpena: sig for the next cycle I think, current cycle is still old rules? 13:38:52 oh, true 13:38:55 yep 13:40:29 o/ 13:40:34 alright, so shall we branch this week, or early next week? 13:40:39 #chair sboyron 13:40:39 Current chairs: dirk hberaud jpena sboyron 13:41:00 I'd propose to do it on Monday 13:41:13 +1 for the next week 13:41:21 hey 13:41:26 #chair toabctl 13:41:27 Current chairs: dirk hberaud jpena sboyron toabctl 13:43:31 jpena: monday works for me 13:43:49 we just need a review I think, any volunteers? 13:44:23 we have a recent core, just saying :P 13:44:30 dirk: let's go 13:44:53 this is a sample review: https://review.opendev.org/#/c/645896/ 13:44:58 ok, I'll do it then 13:44:59 ack 13:45:07 sorry, hberaud will do it then 13:45:27 #action hberaud branching 13:45:27 #action hberaud to send the review for stable/victoria branching on Monday 13:45:32 one more question, do we have pending changes for renderspec and pymod2pkg that we'd like to release 13:46:00 toabctl: can you help us with newer releases of those? I think with the switch to SIG the openstack release team will no longer do releases for us, so we need to recover access to pypi 13:46:17 we have https://review.opendev.org/752272 for renderspec and recent Fedora versions 13:46:58 dirk, so we need to manually create the sdist tarball and push it to pypi? 13:48:01 jpena: ah, the famous py 3.8 breakage. reviewed. 13:48:09 I guess at some point we need to suck it up and switch to "distro" 13:48:15 toabctl: yes, exactly. 13:48:30 urgs :( 13:48:52 toabctl: the discussion was a few weeks ago. basically the "offical project status" requires PTL nomination, which we failed to comply to last time, so evrardjp suggested to become a SIG instead, where we don't have to follow that 13:49:00 the downside is that the openstack release team no longer cares about us 13:50:25 and we are no longer listed as official project? 13:50:28 ideally we'd build a release pipeline elsewhere (github?) 13:50:49 afk for 2 min... 13:51:08 toabctl: my understanding is that SIGS sare still "official" part of openstack, just not a openstack project anymore 13:51:25 as we never "released" with openstack, I think thats kind of fine 13:52:32 as long as we can stay inside the openstack/ namespace I guess it's fine 13:53:04 and yes, I can do the releases on pypi 13:53:16 but also please let me know your pypi user so I can add you to the project, too 13:53:39 what do we releases on pypi ? renderspec only ? 13:53:47 and pymod2pkg 13:53:51 ok thx 13:55:00 can't we host these projects under the independent release model? 13:55:29 I've no much experience with SIGs 13:56:27 hberaud: what do you refer to with "independent release model"? I'm not familiar with that 13:56:47 dirk: https://releases.openstack.org/reference/release_models.html#independent 13:56:50 I think it's what we currently have/had 13:57:31 my understanding is, that we can't use *any* release model as a SIG. now we have to do it on our own. right? 13:57:45 toabctl: ack 13:58:05 toabctl: yes, thats how I understood it 13:58:20 indeed renderspec is already an independent deliverable https://github.com/openstack/releases/blob/master/deliverables/_independent/renderspec.yaml 13:58:37 And this model imply to have a PTL ? 13:59:43 it's still official, it's not _released_ with the deliverables 13:59:43 however, it's not a big deal to manage our own pypi releasing 14:00:03 because most of the deliverables were independent it didn't matter much 14:00:16 is there anything preventing us from reusing Zuul jobs for that? 14:00:17 next to that toabctl already mentioned the willingness for more independence on how to tag 14:00:22 no you can! 14:00:33 the pipelines still exist 14:00:45 evrardjp, but I still like all the automation arround it. this now feels more like a step back imo 14:00:53 (as the release pipeline is an openstack wide thing) 14:00:54 you can implement your own jobs 14:01:18 but you said you wanted to tag per commit at some point, right? 14:01:23 instead of the release repo, I guess we could just do "git push gerrit tag ..." after tagging locally, and that should be it 14:01:25 so that was highly inconvenient 14:01:25 yeah. 14:01:33 now you can do that easier 14:01:34 provided we have the right jobs 14:02:03 jpena: correct you can ask to have the rights to push tags in gerrit. You can also do that immediately in gerrit interface 14:02:11 I found the documentation somewhere a few weeks ago. basically we need to create a rpm-packaging-release-team gerrit group 14:02:15 but I thought that you might want to have your own way, due to the previous requests 14:02:21 which will get additional privileges (push branches, and tags iirc) 14:02:28 correct 14:02:38 ok, then I think it's not a big deal 14:03:01 and yes, you are still in openstack/ 14:03:05 toabctl: ^ 14:03:08 anyone interested in figuring out how to reuse the release tooling? 14:03:15 Merged openstack/renderspec master: Improve distro detection for recent Fedora releases https://review.opendev.org/752272 14:03:27 I remember that was one of things you were looking for, so I tried to find the best solution 14:03:35 evrardjp: do you happen to know how to end up on openstack.org/sigs/ ? is that done automagically? 14:03:54 isn't it already done? 14:04:06 I think I have a patch for governance-sigs repo 14:04:11 and one for governance 14:04:14 sorry https://governance.openstack.org/sigs/ 14:04:30 evrardjp: ok, so the page is generated from that review? 14:04:41 just a sec 14:04:44 yes 14:05:25 dirk: I already have lot of topics to manage for the next weeks but I can help if needed 14:05:54 https://review.opendev.org/#/c/752659/ 14:05:58 I can also give it a look. We have a similar setup to upload to pypi in some projects using gerrit+zuul 14:06:01 someone else is interested too? 14:06:08 that should be good enough to have the page refreshed 14:07:22 I can have a look on it too 14:07:30 thanks hberaud sboyron 14:07:45 lets have a checkpoint on this then next week, we'll certainly have some news then one way or another :) 14:08:03 I suggest to push a release for renderspec before wallaby though, so that we can figure out the release part without pressure 14:08:10 but I think I'll need some help about understanding the exact need, I'm not yet familiar with all these process here 14:08:11 agreed 14:08:30 dirk, imo for a release, we need to be able to tag it in git 14:08:42 what I hear is that we can't do that currently, right? 14:09:07 FYI I intended to make this merge _after_ the release, so that patch might have been merged too soon, sorry for that 14:09:30 toabctl: that should be easily solvable 14:10:33 toabctl: we're still in victoria cycle, for which we have release team support 14:10:47 toabctl: my understanding is that the change to SIG hasn't happened, it will happen in the wallaby cycle 14:11:01 ah. ok 14:11:32 of course its a kind of nice good by gift to push a few reviews, but I'd hope they're okay with it 14:11:43 its not particularly difficult to review/handle 14:12:03 but things have been merged, so let's see :p 14:12:48 (sorry for the delay in answers too, triple booked) 14:13:00 https://review.opendev.org/755521 14:13:12 renderspec 2.1.0 ^^^ 14:13:25 evrardjp: ah, you're saying the review to remove us from "official" has been merged, but the "create sig" hasn't? 14:13:27 nice 14:13:29 :-/ 14:14:10 https://review.opendev.org/#/c/755522/ for a new renderspec release 14:15:41 it's not to remove you from "official" it's to move from project to sig 14:15:48 both are official 14:15:55 both are in governance 14:16:02 https://review.opendev.org/755524 for a new pymod2pkg release 14:16:02 but yeah that is a mistake on the TC side 14:16:09 evrardjp, maybe you can review both!? :) 14:16:24 I will when I am done with my meetings 14:16:34 thanks 14:16:50 please note that I have abandonned my core duties in releases though ;) 14:17:34 * dirk unabandons evrardjp 14:17:49 * toabctl approves that 14:18:16 * evrardjp don't see why dirk and toabctl do that, at all. 14:19:20 evrardjp: you're saying now that you have a goverment you don't obey our orders anymore? ;) 14:19:35 anyway, away the joking, thanks for all the help evrardjp 14:19:54 dirk, sboyron: please prefer to use toabctl's renderspec patch 14:20:37 I missed the drop of py2.7 14:20:38 dirk: hahaha 14:20:50 dirk: trying my best! 14:23:10 ok, I guess we already moved into open floor some time ago, but just in case 14:23:13 #topic open floor 14:23:20 lol 14:23:22 is there anything else you'd like to discuss? 14:23:37 just a quick update on the suse ci instabilities 14:23:40 nothing on my end 14:23:57 it turns out we have some weird network/dns lookup issue, it is being looked at atm 14:24:18 so in case jobs fail with all builds as "excluded" that means you hit the problem, and a recheck hopefully fixes it 14:24:28 There is several old review still opent on rpm-packaging, shouldn't we abandon these ? 14:24:35 there is a caching squid, and more often than not it stops to do dns resolutions :( 14:25:21 ack dirk, thx for these explanations 14:25:59 * hberaud eject (school run) 14:26:57 sboyron: which one in particular? 14:27:09 dragonflow we could abandon 14:27:36 the singlespec one needs rebase, and the vmware-nsx* ones could theoretically be reviewed 14:28:04 openstack-macros: Drop singlespec file 14:28:15 Populate x/ subdir 14:28:25 Add the networking-l2gw-tempest-plugin package 14:28:26 well, the populate x/ dir is still valid 14:29:09 there were a few projects that we package moved around as part of the opendev project, so now rpm-packaging has stuff in openstack/ subdir that is not in openstack/ subdir on review.opendev.org 14:29:23 so we do have to create the x/ subdir imho and run ci over it as well 14:29:33 I just failed to fix the downstream ci for it 14:32:23 ok thx for these status I'll try to find some time to review/patch them. 14:33:35 hey, we're running out of time. Let's close and continue discussion if needed 14:33:38 #endmeeting