Wednesday, 2020-05-13

*** tetsuro has joined #openstack-release00:30
*** tetsuro_ has joined #openstack-release00:40
*** tetsuro has quit IRC00:43
*** redrobot has quit IRC01:34
*** tetsuro has joined #openstack-release01:44
*** tetsuro_ has quit IRC01:47
*** tetsuro_ has joined #openstack-release02:12
*** tetsuro has quit IRC02:15
*** tetsuro has joined #openstack-release03:06
*** tetsuro_ has quit IRC03:09
*** jovial[m1 has quit IRC03:15
*** dustinc has quit IRC03:34
*** tetsuro_ has joined #openstack-release04:06
*** tetsuro has quit IRC04:09
*** ykarel|away is now known as ykarel04:26
*** evrardjp has quit IRC04:36
*** evrardjp has joined #openstack-release04:36
*** diablo_rojo has quit IRC04:44
*** dmellado has quit IRC04:45
*** udesale has joined #openstack-release05:36
*** jbadiapa has joined #openstack-release05:36
*** tetsuro has joined #openstack-release05:43
*** tetsuro_ has quit IRC05:47
*** ysandeep|away is now known as ysandeep05:50
*** tetsuro_ has joined #openstack-release06:23
*** tetsuro has quit IRC06:26
*** tetsuro has joined #openstack-release06:29
*** tetsuro_ has quit IRC06:32
*** slaweq has joined #openstack-release06:51
*** witek_ has joined #openstack-release06:59
*** vishalmanchanda has joined #openstack-release07:03
*** tetsuro_ has joined #openstack-release07:24
*** tetsuro has quit IRC07:27
*** rpittau|afk is now known as rpittau07:29
*** amoralej|off is now known as amoralej07:34
*** tosky has joined #openstack-release07:35
*** dmellado has joined #openstack-release07:37
*** e0ne has joined #openstack-release08:05
*** AJaeger has joined #openstack-release08:07
AJaegerrelease team,  FYI, I'll start updating docs.o.o to point to Ussuri release in a bit. See
* AJaeger likes to have it up early in case of problems or bugs...08:12
*** ykarel is now known as ykarel|lunch08:30
hberaudAJaeger: thanks for the heads up08:39
*** tetsuro_ has quit IRC08:47
openstackgerritThierry Carrez proposed openstack/releases master: Test release with brand-new mirror-on-tags jobs
ttxsmcginnis, hberaud: if you could +2a ^09:05
hberaudttx: ack09:05
AJaegerttx, is 727665 released?09:06
AJaegermerged I mean09:06
ttxI added a depend...09:06
ttxnot enough?09:06
AJaegerah, depend is enough ;)09:07
ttxhberaud: maybe hold for 2 min09:07
hberaudttx: ack09:07
*** tetsuro has joined #openstack-release09:07
ttxor maybe just +2 it and I'll +a it when the other merges09:07
smcginnisThe dependency is close.09:08
smcginnisAnd its merged.09:08
ttxok it's in, +2a away09:08
AJaegerttx, just saw a github mirror failure:
ttxAJaeger: I see the issue09:10
smcginnisShould we block that release-test patch?09:10
ttxLet me explain09:10
ttxBasically the old Gerrit-wide replication did sync all references, including all refs/changes refs to Gerrit patchsets09:11
ttxThe per-repo job does a straight mirror, so that does not replicate refs/changes09:11
ttxso on the mirror, it does remove all refs/changes on Github09:12
ttxWhich will fail miserably on GitHub if there are too many09:12
ttx(that's the error you got)09:12
ttxTo avoid that...09:12
ttxI did clean up asynchropnously all refs/changes on github09:12
ttxBUT I probably missed some repos09:12
ttxlike openstack-manuals09:13
*** dtantsur|afk is now known as dtantsur09:13
ttxLet me clean it up async now09:13
AJaeger shows a few...09:13
AJaegerjust one page going back 2 weeks...09:14
AJaegerthanks, ttx!09:14
ttxI only cleaned up those in projects.yaml... so those in sigs or tc did not get picked up, which explains the majority of those09:14
ttxI should have a look at the others :)09:15
ttxthe cleanup just takes forever as GitHub does not like when you try to delete more than 100 refs at at time09:17
ttxthe others look like transient fails09:19
AJaegerhope that with all the tags we don't run into API limits on github... But sigs or tc repos won't get tagged ;)09:20
*** tetsuro has quit IRC09:23
*** ykarel|lunch is now known as ykarel09:24
ttxOK confirmed all the others were transient errors. Only governance and openstack-manuals were stuck09:24
ttxCleaning them now09:24
openstackgerritMerged openstack/releases master: Test release with brand-new mirror-on-tags jobs
smcginnisTag complete, mirror job now running.09:33
ttxyep, looks good so far09:33
ttxnew tags up09:34
smcginnisSince there aren't any good ussuri songs:
smcginnislogstash error, but doesn't appear to be important.09:35
ttxI guess one question is whether a failure in mirroring would prevent the rest of the release jobs from running09:35
ttxBut I think those are independent09:35
smcginnisI think so.09:36
ttxOK ,let's wait a couple more minutes and send the torpedo away09:38
smcginnisProbably good to let the rest of the jobs finish, just to be safe.09:38
AJaegerSo, now points to Ussuri as new release. If anybody wants to double check...09:46
smcginnisLooks right to me.09:46
AJaegerthanks - looks good to me as well. Still missing deployment projects but they are not released yet...09:47
smcginnisYeah, might be awhile yet before the cycle-trailing things are ready.09:48
smcginnisttx: OK, I think the important things are done. We can probably go ahead now.09:49
ttxok, anyone else wants to pile up their +2?09:49
AJaegerwould be good to remind them when they are ready to send patches for openstack-manuals to get added to the index09:50
ttxok we are ON10:00
smcginnisBombs away!10:00
fungii'm on hand as well, just in case10:08
*** ysandeep is now known as ysandeep|lunch10:09
smcginnisSo far so good. Still running all jobs in gate.10:09
AJaegermorning, fungi10:11
fungithanks, and a guten morgen to you10:12
smcginnisValidate should be ~20-30 minutes. Better than 3 hours. :)10:12
AJaegerfungi: lunch time now ;) Time to start cooking...10:13
fungiindeed, it must be nearing noon there10:15
fungii bet the sun isn't even up for smcginnis yet10:15
fungiit's only just starting to rise here10:16
smcginnisThe sky is just starting to lighten up.10:23
smcginnisvalidation job wrapping up.10:30
openstackgerritMerged openstack/releases master: Ussuri final releases for cycle-with-rc projects
fungigood so far10:31
hberaudsmcginnis: then I think I can start to check missing-releases10:32
*** udesale_ has joined #openstack-release10:36
smcginnishberaud: Thanks. Just need to wait for the tagging jobs to finish so the expected packages get uploaded. Then we should be all set.10:37
hberaudsmcginnis: ack10:38
*** udesale has quit IRC10:38
hberaudsmcginnis: can you paste the url to follow this job?10:39
fungihberaud: the release pipeline at
hberaudfungi: thx10:41
fungithose are being generated by the one in the release-post pipeline10:41
hberaudyep I seen them10:42
fungithe tag-releases job running for the openstack/releases commit there10:42
smcginnisOn masakari-dashboard, so making good progress.10:42
fungithe console stream for that tag-releases build is
fungilooks like cache_remote is trying to use the old /opt/git path still, and spits out a warning10:50
fungiWARNING: Cache directory /opt/git/openstack/nova does not exist, ignoring.10:51
smcginnisfungi: Is that where it tries to use a local existing clone to speed up the current clone?10:51
fungii believe so10:51
smcginnisI remember seeing that before. I wasn't aware there were changes.10:51
smcginnisWhat is the new path?10:52
smcginnisfungi: Looks like our script defaults to that path if ZUUL_CACHE_DIR is not set.10:53
fungior maybe that's not obsolete, but it certainly doesn't seem to be finding it10:53
*** ysandeep|lunch is now known as ysandeep10:55
fungilooks like it should be using /opt/git/
fungiinstead of /opt/git/openstack/nova10:56
smcginnisAh, additional namespacing for more tenants probably?10:57
fungiwell, for more potential source connection names10:57
openstackgerritSean McGinnis proposed openstack/releases master: Use updated git cache dir default path
smcginnisTagging almost complete.11:00
*** rpittau is now known as rpittau|bbl11:04
funginow it's a matter of waiting for the release pipeline to empty out and checking for any job failures11:05
ttxlooking good so far11:10
openstackgerritSean McGinnis proposed openstack/releases master: Use updated git cache dir default path
smcginnisLooks like only email announcements and update constraints jobs left.11:17
openstackgerritDmitriy Rabotyagov (noonedeadpunk) proposed openstack/releases master: Release OpenStack-Ansible Stein
funginothing new at since friday11:20
openstackgerritDmitriy Rabotyagov (noonedeadpunk) proposed openstack/releases master: Release OpenStack-Ansible Train
smcginnistacker-horizon announce-release failure.11:23
hberaud`missing-release` return a huge list (922 errors) or I missed something =>
fungiyep. just showed up11:24
smcginnisLogs make it look like it worked though.11:24
fungiyep, just log collection failed11:26
fungirsync hit an ssh socket timeout11:26
smcginnisSafe to ignore then.11:26
smcginnishberaud: I see a lot of RC version in that list. Did you pull the latest from master before running?11:26
smcginnisRunning locally now too. I am also getting a lot of missing tarballs and signatures.11:27
smcginnisfungi: Is there a delay in syncing tarballs.o.o?11:27
hberaudsmcginnis: normally yes, I recheck11:27
fungismcginnis: volume release occurs every 5 minutes for that site, but may take longer depending on the amount of data... i'll check on its progress11:28
smcginnisOh, I think we can drop the check for python 2 and python 2/3 wheels.11:28
hberaudsmcginnis: ah wait...I point on 27c15e1028c3a0c097baac273245284d07c5d87111:28
hberaudsmcginnis: I thought to be up-to-date...11:29
smcginnisI've done that many times. ;)11:31
fungismcginnis: looks like the tarballs site volume should be in sync, what's missing?11:31
fungii see a nova-21.0.0-py3-none-any.whl and nova-21.0.0.tar.gz and associated .asc files at
ttxexample: missing11:39
ttxbut it's not... hmm11:39
fungiremember that's a redirect now11:40
ttxyeah the script seems a bit out of date11:40
fungimostly making sure the script successfully follows permanent redirects11:40
hberaudgood point11:42
ttxmissing = ((response.status_code // 100 != 2))11:42
openstackgerritTobias Urdin proposed openstack/releases master: Release puppet-ceph 3.1.1
smcginnisSorry, I've been updating the script. I think I might have it.11:43
smcginnisWe also need to drop checks for py2 wheels.11:43
ttxI'll run the W3C link checker in parallel11:44
openstackgerritSean McGinnis proposed openstack/releases master: Update missing cmd to follow redirects
ttxsmcginnis: can you paste the report you get with this version?11:50
hberaudlooks better now :)11:52
smcginnisYep, almost done.11:52
smcginnisLooking a LOT better than it had been. :)11:52
hberaudI agree :)11:53
smcginnisBarbican one is indeed missing a whl.11:54
smcginnis10.0.0 only has the tar.gz file and signature.11:55
smcginnisceilometermiddleware appears to not have completed the drop-py2 goal, so there is a py2.py3 wheel.11:55
smcginnisI can update the script to still check for that. That may eliminate a few of these.11:56
hberaudgood idea11:57
*** amoralej is now known as amoralej|lunch11:58
AJaegerceilometermiddleware merged in November11:58
smcginnisfungi: Do you know - are py2.py3 wheels built if the universal flag is set? Or some other trigger?12:01
fungithat's what typically causes it to happen, pretty sure. i'll double-check that project12:02
AJaegerI don't see universal flag set recently12:02
AJaegerI'm checking the repo right now and do a few cleanups12:04
AJaegerdidn't see anything obvious ;(12:04
smcginnisI had expected to see universal=true in their setup.cfg, but it's not there.12:04
smcginnisSo not sure why 2/3 wheel would be created instead of just 3.12:05
fungitheir 2.0.0 release was in december12:06
smcginnisHmm, script returned happy, but I would have expected an error at least for barbican.12:08
fungii don't see evidence that ceilometermiddleware ever set universal in their setup.cfg, in fact, looking through git history12:11
smcginnisOh, sorting threw me off for barbican. They do have a 2/3 wheel.12:13
smcginnisRunning the script one more time after some updates.12:13
fungiyeah, ceilometermiddleware never even had a [wheel] section in their setup.cfg at all12:13
smcginnisBut I think we are in good shape.12:13
AJaegerbarbican took ages to get drop-2.7 changes in, that repo was in a sad state12:14
smcginnisI do see it for barbican:
smcginnisI wonder what could be triggering that for ceilometermiddleware12:14
AJaegerdid we change our tooling after that release?12:16
fungiunfortunately, with their release happening in early december, we no longer have the job log from it12:16
fungiit's been over 6 months12:16
AJaegerI41de5a1f57eed0a165bcd93ab1585003c3d478fb merged in January12:16
AJaegerthat's it, I guess12:16
AJaegersmcginnis: "Stop forcing universal wheel releases"12:16
*** ysandeep is now known as ysandeep|brb12:17
fungioh, yep, that was it12:17
fungigood memory, AJaegermystery solved12:17
AJaegersmcginnis: so we would need to re-release ceilometermiddleware or timetravel ;)12:17
smcginnisOK, that makes sense. Thanks AJaeger.12:18
smcginnisThat one is a cycle-with-intermediary deliverable, and they had just the one release in December. Before that job change.12:18
hberaudmake sense12:19
smcginnisAnd check with current script is happy again.12:19
openstackgerritSean McGinnis proposed openstack/releases master: Update missing cmd to follow redirects
fungiokay, so any outstanding unknowns from the missing report?12:19
*** rpittau|bbl is now known as rpittau12:19
smcginnisNone at all. It appears to be happy with everything.12:20
AJaegereverything? Wow!12:20
smcginnisI will send out the announcements. Looks like we are all set.12:20
fungithanks! smooth as glass12:20
ttxhmm just a sec12:21
fungittx: any unexpected issues raised by the linkcheck?12:21
smcginnisFeel free to check my work with
ttxthe linkscheck does not like us now that our page is too big.12:21
ttxsmcginnis: announcement should be sent closer to 15utc12:22
ttxwe need to set release as released first12:22
ttxor did I miss that step12:23
smcginnisNope, I'll get that now.12:23
ttxok :) Was worried about "sending announcements" :)12:24
smcginnisYeah, getting a little ahead of myself.12:24
*** tkajinam has quit IRC12:24
ttxI'm crossing lines on
openstackgerritSean McGinnis proposed openstack/releases master: Mark Ussuri as released
ttxall the tags were successfully pushed to GitHub, fwiw12:28
ttxfungi: we noted this morning that tags were actually not synced12:28
ttxsince the mirroring job only ran in post12:29
fungiright, i warned about that back when the job was implemented12:29
fungifolks didn't seem concerned at the time12:29
ttxI should write down warnings12:29
fungi"tags will show up the next time a change is merged" was the answer, so i figured it wasn't a big deal ;)12:30
AJaegerfungi, we merged a change to openstack-zuul-jobs to sync after release12:30
fungiAJaeger: i saw, good thinking12:30
smcginnisIt would be eventually consistent.12:30
smcginnisJust the value of "eventually" could be quite large for some repos.12:31
fungii still think that job could just push the triggering branch or tag, rather than doing a full mirror on every run, but more optimization12:31
smcginnisttx: Can you think of any reason we have steps 4 and 5 called out as two separate steps in
ttxit comes with an ominous warning note12:33
ttx"Only workflow when previous step *fully* ready"12:33
ttxso I think we should honor whoever wrote that warning12:34
smcginnisThat's what I am wondering about. I don't see any reason why that should be the case.12:34
ttxok, let me dive into it12:34
ttxthere is a lot riding on that default series value12:35
openstackgerritSean McGinnis proposed openstack/releases master: Update release week process
smcginnisWell, assuming we don't have a real reason for the separation, that will be ready. ^12:37
hberaudsound like a warning at the entrance of a pyramid that warns of calamities incurred by profaners.12:38
ttxThat note was added (by you) after the Stein release, referencing "changes in process found during the stein release process at Denver PTG"12:38
fungiabandon all hope, ye who enter here12:39
smcginnisttx: Just saw that.12:39
ttxsmcginnis: at that time you also collapsed the patch into one patch12:39
ttxI'm pretttty sure we added that note for a good reason then12:39
smcginnisI wonder if there was another step we were thinking about after the discussion with changing auto releases and things like that.12:39
ttxIf it had worked correctly you probably would not have added that12:39
smcginnisI combined them for train too.12:40
fungimaybe it was because you realized a tag hadn't been pushed for something and needed to be added at the last moment?12:41
fungicould dive into irc logs from that time12:41
smcginnisI fixed up bulleting in that patch. Since it was a separate paragraph, I probably just threw it into it's own section. I really can't recall, but I can't think of any reason now why we would need to break it out.12:41
smcginnisfungi: Was just thinking of finding the timestamp of when that patch was orginally submitted. ;)12:42
smcginnisNo comments in IRC around the time.12:44
ttxyeah, we were in person12:44
smcginnisLooking through for clues.12:45
smcginnisWas done interactive collab style:
ttxyellow, what were you thinking12:46
smcginnisAssuming "ready" is approved and merging, and since we say this one can be approved and be staged on top of the previous one, the only separation here would be if we need to do another check of anything before approval. There's nothing else to do, especially since the previous step is just updating the docs.12:48
ttxyeah, let's live dangerously12:49
smcginnisI better shower quick before meetings start up. The "mark released" patch has zuul +1. I will be back soon.12:49
ttxThis was far too calm12:49
smcginnisNo kidding. I think this was the calmest I've been a part of so far.12:49
ttxwe should change it again12:49
ttxwe are getting soft12:49
smcginnisWhen is zuul v4 coming out?12:50
*** ianychoi has quit IRC12:55
*** amoralej|lunch is now known as amoralej13:00
fungismcginnis: not too long actually13:08
fungiv4 and v5 are both planned13:08
smcginnisLet's hope, for our piece at least, that migration goes much smoother than the transition to v3 went.13:08
fungibut the major version bumps aren't for significant user-facing changes, they're for admin-facing changes13:08
smcginnisThat's good.13:08
fungiservice architecture improvements for statefulness and high availability13:09
*** ykarel is now known as ykarel|afk13:09
openstackgerritSean McGinnis proposed openstack/releases master: Mark Ussuri as released
ttxI'll W+! this ^13:49
*** ykarel|afk is now known as ykarel13:59
ttxcommunity meeting on Ussuri release starting now14:00
smcginnisttx: Good to send the announcement now?14:00
ttxsmcginnis: yeah, it's fair game now14:01
ttxjust wait until is processed14:01
ttxso that it shows as released on
ttxsmcginnis: ^14:02
openstackgerritMerged openstack/releases master: Mark Ussuri as released
diablo_rojo_phonWhoop whoop!14:23
ttxreleases.o.o page still not updated14:33
smcginnisJob is done, but I'm not seeing the updates.14:35
smcginnisMaybe there is another sync step now. I think it was stated that happens every 5 minutes, so will keep checking.14:35
ttxProbably waiting for the refresh14:35
ttx cc: fungi14:35
smcginnisIt's updated now.14:37
fungistanding down14:37
smcginnisI noticed releases.o.o doesn't have a favicon. Anyone know if that is an openstackdocstheme setting or something?14:37
smcginnisttx: Sending announcements now.14:38
*** ysandeep|brb is now known as ysandeep14:38
smcginnisNeed moderator approval for openstack-announce.14:38
fungiand yeah, it'll generally get a vos release batch run every minute evenly divisible by 5, but the volumes released together are processed serially so, for example, a large amount of documentation churn could cause the docs.o.o release to take a couple of minutes before it gets to the releases.o.o volume14:39
fungii can get the announce moderation queue14:39
*** redrobot has joined #openstack-release14:43
ttxnow back at paying attention to the community meeting14:43
mnaserttx, smcginnis: can i technically say "ussuri is out" at the end of the call?14:45
ttxmnaser: yes, it's hit the ML14:45
ttxIt's out now if you want to mention it between speakers14:45
smcginnismnaser: Yep!
mnaserill wait till the end :P14:46
smcginnisI feel like someone behind the scenes preparing for the big keynote reveal announcement. :)14:47
mnaserahaha :p14:47
ttxsmcginnis: welcome to marketing14:47
ttxsmcginnis: congrats on yet another on-time release, fearless leader!14:57
smcginnisYou too.14:57
smcginnisGreat job team!14:57
smcginnisOf course I couldn't make it without at least one typo in an email.14:58
*** priteau has joined #openstack-release15:02
clarkbsmcginnis: that is how you let people know its handcrafted open source software. Not factory produced. Just don't let Zuul in on it :)15:16
smcginnisThat's it - it was on purpose to be authenticate. Yep, totally. :D15:17
*** zxiiro has joined #openstack-release15:32
*** ykarel is now known as ykarel|away15:35
*** dtantsur is now known as dtantsur|afk15:36
openstackgerritSean McGinnis proposed openstack/releases master: Add release note links for newly added pages
smcginnisttx: Still around? I think we want to push that through right away.  ^15:45
smcginnisOr hberaud15:45
hberaudsmcginnis: this patch kill my browser :)15:46
hberaudsmcginnis: this is something that we want => ?15:57
hberaudsmcginnis: I mean the empty reno15:57
hberaudI guess yes to prepare the floor for ussuri...15:58
smcginnishberaud: Yeah. If the landing page is there, then we want to add it. Hopefully eventually they will actually have content in there.15:59
*** AJaeger has left #openstack-release16:00
*** hberaud has quit IRC16:13
*** hberaud has joined #openstack-release16:13
*** witek_ has quit IRC16:15
openstackgerritMerged openstack/releases master: Add release note links for newly added pages
*** rpittau is now known as rpittau|afk16:21
*** camelCaser has quit IRC16:22
*** ccamel has joined #openstack-release16:22
openstackgerritAndreas Jaeger proposed openstack/releases master: Release openstackdocstheme 2.1.0
*** AJaeger has joined #openstack-release16:24
AJaegersmcginnis: hope that docstheme release is not too quick ;)16:24
smcginnisAJaeger: You had that queued up and waiting, didn't you. :D16:25
AJaegersmcginnis: I was waiting for the merges first...16:25
smcginnisGood to get that out. Some good stuff has merged.16:25
AJaegerexactly - and melwitt wants parallel releasenotes building (I saw nova building for 1 h in Zuul)16:26
*** witek_ has joined #openstack-release16:26
smcginnisHopefully that helps. I think we need to get all repos to raise their openstackdocstheme lower constraint to make sure we can apply that community-wide in project-config though. Or else drop the -W flag.16:27
smcginnisThe other issue there was it rebuilds the docs for each translation. Not sure if we can speed anything up with how that is done. Probably not since it really does need a full build of the docs to be done per language.16:28
AJaegeryeah, indeed, we need to update everywhere - but we're at the beginning of a cycle ;)16:29
melwittyeah. I wanted to ask the zuul ppl if there's a way we could parallelize that task, so that each language build process gets kicked off at the same time and might benefit from being able to run in parallel?16:29
smcginnisYes, hopefully enough time to get it all in.16:29
AJaegerwe really need the translated docs rebuild everytime, otherwise content is missing16:29
smcginnisThat's a good idea melwitt.16:30
AJaegermelwitt: the job is in zuul-jobs if you wnat to have a look.16:30
melwittI'm not sure how much '-j auto' will help us but I'm trying it out to see what's the speedup. note that we'd also need a new reno release because it too added parallel_read_safe recently. I dunno if a release is already proposed yet or not16:31
AJaegermelwitt: yes, we need both reno and docstheme16:31
smcginnisIt's been queued up, but we didn't want to release a new reno until ussuri was out just to be safe.16:31
smcginnisShould be able to get that out now too.16:31
melwittAJaeger: thanks! I am aware of that but I wasn't sure if/how a section/label can be prescribed to run in parallel. because "something" is running that label serially somewhere. I will try to find where and also ask if I can't find16:32
melwittok cool16:32
smcginnisIn fact, hberaud, would you mind taking a look at
*** udesale_ has quit IRC16:33
melwitt*serially, one run per language16:33
AJaegermelwitt: adding -j auto in is the best way IMHO. Otherwise: You need to rewrite that completely16:33
hberaudsmcginnis: done16:34
melwittAJaeger: ok, good to know. I will continue to focus on trying that out and see the gain, then go from there. thanks16:34
AJaegermelwitt you can run that script locally to test16:34
smcginnisThat is still serially per language. It would be nice to make that part parallel.16:35
smcginnisBut not sure of a good way to do that when running a bash script.16:35
smcginnisAt least the builds should take less time, so sooner to get to the next language.16:35
* melwitt nods16:35
*** evrardjp has quit IRC16:36
*** evrardjp has joined #openstack-release16:36
AJaegerif you have another script, could be python, to do the building, we can use that...16:37
smcginnisIt may not be worth it. We only have a limited number of cores in the VM that runs the job, so we risk adding too much contention with all of this parallelism.16:38
AJaegerso, let's try the sphinx -j auto16:39
smcginnisThinking it through, that's probably the best option.16:39
AJaegerbut first release reno and openstackdocstheme - and update a 'few' repos16:41
*** ysandeep is now known as ysandeep|away17:12
openstackgerritMerged openstack/releases master: Release openstackdocstheme 2.1.0
*** amoralej is now known as amoralej|off17:18
*** priteau has quit IRC17:38
*** e0ne has quit IRC17:55
*** jbadiapa has quit IRC19:37
melwittI kicked off a build after the openstackdocstheme release merged but it's still saying "WARNING: the openstackdocstheme extension does not declare if it is safe for parallel reading, assuming it isn't - please ask the extension author to check and make it explicit" is there something else that needs to be done to get the latest version?19:50
AJaegermelwitt: local build?19:52
AJaegerare you really using openstackdocstheme 2.1.0?19:52
melwittno CI build. I thought after the release the new version would get pulled in? sorry I don't know anything19:57
AJaegermelwitt: if you build locally with constraints, the constraints might limit to an older version19:58
AJaegermelwitt: install the theme into your venv/tox and check with pip freeze that it's really there19:59
melwittI'm just building in upstream zuul, not locally. maybe I need to add the version to test-requirements.txt? I'm just using DNM patches19:59
melwittupstream zuul job does a completely different thing than our local tox.ini run20:00
AJaegermelwitt: have a link for me?20:00
AJaegerupstream and local both use constraints the same way20:01
melwittAJaeger: I don't really want to show you :) but it's this. ignore the actual change bc it's a dummy, it's the Depends-On where things are really happening
melwittthe releasenotes CI job does _not_ run our tox -e releasenotes target at all20:02
AJaegermelwitt: correct, CI does not use tox -e releasenotes since we build the translated docs centrally20:03
AJaegeryeah confusing ;(20:03
melwittso now I'm thinking if I add the version to test-requirements.txt and lower-constraints.txt in my DNM nova patch maybe that will do it?20:03
melwittI'm just testing things out to see the job run time effect the -j auto will have20:04
melwittwe're currently taking around an hour to run the job20:04
AJaegergood idea20:05
AJaegerI see the job running but can't find the version of openstackdocstheme yet ;(20:05
melwittok cool thanks :) this is my first time working with any docs related jobs, lots of new to me things20:06
melwittyeah it doesn't even say it in the job output, which surprised me20:06
melwitt(the openstackdocstheme version)20:06
melwittlooking at an older run, it doesn't seem to mention the version anywhere
AJaegeradd a "pip freeze " to roles/build-releasenotes/tasks/main.yaml ;)20:10
AJaegerthe pre-playbook installs sphinx and shows the version on that one20:11
AJaegeryou'll get that output at the end of the run20:12
AJaegermelwitt: that output is only in the JSON, not in the console20:13
AJaegerso, please wait a bit longer...20:13
melwittdang I see now on the old run20:13
melwittI had been looking at previous runs and didn't know the json could have more info in it20:14
AJaegerI'll have to leave now, will read tomorrow what you figured out and answer any more questions.20:14
smcginnismelwitt: You'll need yet another depends-on I'm afraid:
AJaegersmcginnis: yes, that's most likely the problem20:14
AJaegerbye for today!20:14
smcginnisGood night AJaeger20:15
melwittsmcginnis: ah thanks. np, I was going to try test-requirements anyway but would've been blocked by that :P20:15
melwittso was planning another rev anyhow20:15
melwittthanks AJaeger!20:15
melwittkewls depends-on that works to pull in the new openstackdocstheme version. api-ref job already finished so I could peek20:30
*** vishalmanchanda has quit IRC20:31
smcginnisLooks like the release notes job is just getting going.20:31
melwittyeah, I'm glad it got queued up so quickly this time. nice20:32
smcginnisThat was quick. I saw a few patches that took nearly two hours to kick off.20:32
melwittyeah, previous to this PS it was taking hours before jobs started. this time it was instant. I wonder if depends-on openstack/requirements gave it a higher priority20:34
smcginnisThese scanning steps of reno seem to really take a long time for some of the older branches.20:35
smcginnisBut it starts at ocata, which is still Extended Maintenance, so not sure that could be made static.20:36
smcginnispike took 1:15 to scan release notes.20:36
smcginnisSo if all the extended maintenance branch release note pages were made static, maybe with a note in there to instruct people what to do manually if new releases notes needed to be added, that looks like it could cut off maybe ~5-7 minutes per build. Time the number of translated builds it needs to do, that could add up.20:39
smcginnisNot sure if it would add up enough to stop timeout problems, but at least buy a little more time.20:39
* melwitt nods20:40
melwittuh oh, looks like our docs job fails with the new openstackdocstheme. guh20:42
smcginnismelwitt: Looks like that failed on the pdf portion.20:54
melwittyeah, I'm looking at it. not yet understanding it tho20:55
smcginnisI wonder if it could be a side effect of
melwittyeah maybe.. gonna try running locally and see if I can get more info21:02
melwittbah turns out my depends-on the reno change didn't work "WARNING: the reno.sphinxext extension is not safe for parallel reading" on the latest DNM patch run21:33
smcginnisHmm. Well, hopefully we will have that out and upper-constraints raised soon.21:42
*** slaweq has quit IRC21:57
*** slaweq has joined #openstack-release22:08
*** slaweq has quit IRC22:13
*** slaweq has joined #openstack-release22:23
*** slaweq has quit IRC22:28
*** tkajinam has joined #openstack-release22:46
*** hongbin has joined #openstack-release23:05
*** tosky has quit IRC23:06

Generated by 2.17.2 by Marius Gedminas - find it at!