Tuesday, 2019-04-02

*** timburke has joined #openstack-release00:00
*** mgoddard has quit IRC00:06
*** mgoddard has joined #openstack-release00:22
*** hongbin has joined #openstack-release00:37
*** dangtrinhnt has quit IRC01:24
*** ricolin has joined #openstack-release01:29
*** dangtrinhnt has joined #openstack-release01:31
*** whoami-rajat has joined #openstack-release01:31
*** ekcs has quit IRC01:46
*** lbragstad has joined #openstack-release02:02
*** bobh has joined #openstack-release02:07
*** bobh has quit IRC02:17
*** ekcs has joined #openstack-release02:30
*** nhbinh_ has quit IRC02:49
*** nhbinh_ has joined #openstack-release02:49
*** hongbin has quit IRC02:57
*** hongbin has joined #openstack-release02:58
*** hongbin has quit IRC03:12
*** hongbin has joined #openstack-release03:14
*** samueldmq has quit IRC03:15
*** ykarel|away has joined #openstack-release03:27
*** hongbin has quit IRC03:31
*** spsurya has joined #openstack-release03:31
*** ekcs has quit IRC03:36
*** ekcs has joined #openstack-release03:36
*** ykarel|away is now known as ykarel03:40
*** ekcs has quit IRC03:45
*** ykarel has quit IRC04:11
*** udesale has joined #openstack-release04:11
*** ykarel has joined #openstack-release04:30
*** lbragstad has quit IRC05:33
*** pcaruana has joined #openstack-release05:35
*** nhbinh_ has quit IRC05:42
*** nhbinh_ has joined #openstack-release05:42
*** ifat_afek has joined #openstack-release05:44
*** jtomasek has joined #openstack-release06:29
*** hberaud|gone is now known as hberaud07:01
*** tesseract has joined #openstack-release07:02
*** rakhmerov has joined #openstack-release07:08
*** tosky has joined #openstack-release07:09
*** e0ne has joined #openstack-release07:16
*** jpich has joined #openstack-release07:38
*** zul has quit IRC07:44
*** ykarel is now known as ykarel|lunch07:58
*** e0ne has quit IRC08:01
*** e0ne has joined #openstack-release08:08
*** openstackgerrit has joined #openstack-release08:15
openstackgerritIfat Afek proposed openstack/releases master: Release Vitrage deliverables for Stein  https://review.openstack.org/64842008:15
*** zbr|pto is now known as zbr08:23
*** cdent has joined #openstack-release08:29
*** priteau has joined #openstack-release08:34
*** ykarel|lunch is now known as ykarel08:50
*** e0ne has quit IRC09:07
*** e0ne has joined #openstack-release09:15
*** dtantsur|afk is now known as dtantsur09:33
*** electrofelix has joined #openstack-release09:34
*** Sundar has joined #openstack-release09:43
Sundarclarkb and all: I followed https://docs.openstack.org/infra/manual/drivers.html#merge-commits to merge the master onto my feature branch. When I do 'git review -R feature/cyborg-nova-pilot', I get the error: "you are not allowed to upload merges". This seems like a permission issue. Who sets the permissions?09:48
SundarMore details at: http://paste.openstack.org/show/748710/09:48
SundarPrevious discussion at: http://eavesdrop.openstack.org/irclogs/%23openstack-release/%23openstack-release.2019-04-01.log.html#t2019-04-01T17:07:2609:50
*** e0ne has quit IRC09:51
*** hberaud is now known as hberaud|school-r09:55
*** zigo has joined #openstack-release10:04
*** priteau has quit IRC10:07
*** priteau has joined #openstack-release10:12
*** e0ne has joined #openstack-release10:28
*** e0ne has quit IRC10:45
*** amoralej is now known as amoralej|lunch10:47
*** diablo_rojo has joined #openstack-release10:51
*** e0ne has joined #openstack-release10:52
*** diablo_rojo has quit IRC11:10
*** udesale has quit IRC11:10
*** hberaud|school-r is now known as hberaud11:13
*** amoralej|lunch is now known as amoralej11:29
*** cdent has quit IRC11:36
*** e0ne has quit IRC11:56
dhellmannSundar : you'll have better luck contacting the infra team in #openstack-infra12:02
*** e0ne has joined #openstack-release12:03
*** diablo_rojo has joined #openstack-release12:16
*** e0ne has quit IRC12:31
*** ifat_afek has quit IRC12:41
*** e0ne has joined #openstack-release12:44
*** mriedem has joined #openstack-release12:52
*** cdent has joined #openstack-release12:55
openstackgerritLuka Peschke proposed openstack/releases master: Release cloudkitty stein rc2  https://review.openstack.org/64931313:07
*** mriedem has left #openstack-release13:08
*** mriedem has joined #openstack-release13:09
*** lbragstad has joined #openstack-release13:22
openstackgerritLuka Peschke proposed openstack/releases master: Release cloudkitty stein rc2  https://review.openstack.org/64931313:25
*** priteau has quit IRC13:28
*** trident has quit IRC13:30
*** trident has joined #openstack-release13:33
*** e0ne has quit IRC13:36
*** e0ne has joined #openstack-release13:42
*** e0ne has quit IRC13:43
*** dikonoor has joined #openstack-release13:59
*** e0ne has joined #openstack-release14:10
*** smcginnis_pto is now known as smcginnis14:16
*** mlavalle has joined #openstack-release14:22
*** hberaud is now known as hberaud|school-r14:24
*** hberaud|school-r is now known as hberaud14:34
*** cdent has quit IRC14:37
Sundardhellman: Thanks14:41
openstackgerritAlex Schultz proposed openstack/releases master: Switch ansible-trole-container-registry to independant  https://review.openstack.org/64934914:42
openstackgerritAlex Schultz proposed openstack/releases master: Switch ansible-role-openstack-operations to independent  https://review.openstack.org/64935014:45
openstackgerritAlex Schultz proposed openstack/releases master: Switch ansible-role-container-registry to independent  https://review.openstack.org/64934914:47
openstackgerritMohammed Naser proposed openstack/releases master: Branch Stein in OSA  https://review.openstack.org/64894714:54
*** e0ne has quit IRC15:11
evrardjpit seems we have issues with docs building with the new sphinx. I will have a look after my meetings of today15:14
smcginnisevrardjp: Did you see the ML thread about the changes with sphinx 2.0?15:15
evrardjpnot yet, but I supposed that was it.15:15
evrardjpI will have a look when $meeting ended I guess :)15:16
smcginnisLooks like we need a new release of sphinxcontib.datatemplates.15:16
smcginnisOr pin to sphinx < 2.015:16
smcginnisI'm catching up yet, but I'll try to look into the status of that too.15:16
*** ykarel is now known as ykarel|away15:18
*** ykarel|away has quit IRC15:23
*** diablo_rojo has quit IRC15:26
*** e0ne has joined #openstack-release15:38
smcginnisevrardjp, dhellmann: https://github.com/sphinx-contrib/datatemplates/issues/615:51
evrardjpI think due to final release being very close it's maybe wise to pin <215:53
evrardjpbut I don't know yet the scale of the changes, so ignore this if not relevant15:53
evrardjpsmcginnis: aren't you on holidays though?15:53
smcginnisevrardjp: No, just a long weekend. I was orginally going to take the whole week off, but decided to save it.15:53
smcginnisevrardjp: We don't follow requirements contraints in our release jobs, so we can update that without an issue.15:54
smcginnisNot sure how many other projects are impacted, so they may want to pin for now. But it's a fairly easy fix once you find out where it's coming from.15:54
evrardjpoh I see15:54
evrardjpI guess we can take popcorn and see if something happens in terms of pinning in the next days15:55
evrardjpin the meantime it's good we are masters of our own destiny15:55
smcginnisIt does help in times like this. :)15:56
dhellmannsmcginnis , evrardjp : in the past we applied the cap for sphinx in the job definition16:05
smcginnisdhellmann: Would you rather cap things?16:06
dhellmannI'm doing a release right now, but we keep playing whack-a-mole with this16:07
dhellmannthey did at least signal the breaking changes this time with a 2.0 version  number :-)16:07
smcginnisYeah, and it was deprecated for awhile.16:08
smcginnisI had done a few cleanups in our (openstack) code, but I didn't think to look at some of the libs we use.16:08
dhellmannnor i16:13
dhellmannthe datatemplates 0.2.0 release should deal with it16:13
dhellmannthanks for the patch :-)16:13
smcginnisGreat, thanks dhellmann!16:13
*** ykarel|away has joined #openstack-release16:13
evrardjpthanks dhellmann -- what would we do without you!16:15
evrardjpsmcginnis: do you have a patch to change the requirements on which I can vote?16:15
evrardjpOSA crew is very thrilled of making stein happen, as you acn see: https://review.openstack.org/#/c/648947/ :)16:16
evrardjpit's amazing!16:16
*** dtantsur is now known as dtantsur|afk16:24
*** e0ne has quit IRC16:28
*** hberaud is now known as hberaud|gone16:46
*** dikonoor has quit IRC16:55
*** jpich has quit IRC16:58
*** ricolin has quit IRC17:17
*** amoralej is now known as amoralej|off17:28
*** ekcs has joined #openstack-release17:45
*** armax has joined #openstack-release17:53
*** e0ne has joined #openstack-release18:04
*** cdent has joined #openstack-release18:09
*** electrofelix has quit IRC18:18
*** cdent has quit IRC18:30
*** spsurya has quit IRC18:32
*** ykarel|away has quit IRC18:40
*** gmann is now known as gmann_afk19:04
*** tesseract has quit IRC19:04
prometheanfirehttps://pypi.org/project/PyECLib/ has not had 1.6.0 pushed to pypi19:12
prometheanfirenot sure if it's managed, bot doesn't show it I think19:13
smcginnisprometheanfire: Who uses that? Swift?19:13
prometheanfirehttps://git.openstack.org/cgit/openstack/pyeclib/ tagged by tim burke19:14
prometheanfirelooks like unmanaged by releases19:14
smcginnisYeah, not listed in the official governance list.19:14
smcginnistimburke: If you're around, friendly reminder in case you need that released. ^^19:15
prometheanfireI pined him in -swift19:15
prometheanfiredidn't check for nick in here like I should have19:15
smcginnisToo many places.19:15
tdasilvawe noticed this issue with pyeclib: http://logs.openstack.org/54/54eaab52d4558a630da3aa32ab5a8439f9e56ce1/release/release-openstack-python/621e584/job-output.txt.gz#_2019-04-01_16_00_54_48242319:16
tdasilvaI think that's the reason it is not being pused to pypi automatically?? but not sure yet how to get around the problem...19:17
smcginnistdasilva: Yeah, from that log it looks like pypi is rejecting it.19:17
smcginnisWhich is odd since that specific tag is an example in the pep: https://www.python.org/dev/peps/pep-0425/#overview19:18
*** e0ne has quit IRC19:20
prometheanfireI've tried to ping heat about missing releases but no response, email thread sound good as next step?19:22
smcginnisYeah, probably.19:23
prometheanfiresmcginnis: you said the heat issue was something to do with a rename somewhere?19:26
smcginnisIIRC, the package name was changed in setup.cfg because it needed to be openstack-heat to be an available name on pypi.19:26
smcginnisThat was done in master (stein) but not in the older stable branches.19:26
prometheanfirethe tarball for 12.0.0-rc1 is missing too19:28
prometheanfirewhich is stein19:28
smcginnisHmm, it uploaded to pypi: https://pypi.org/project/openstack-heat/19:29
prometheanfiremail sent19:30
smcginnisMaybe once fungi is out of his meeting he can let us know if there were any known tarball upload issues.19:30
prometheanfireya, may be more limited to that19:30
prometheanfirehttps://tarballs.openstack.org/openstack-heat doesn't exist19:31
prometheanfirehttps://pypi.org/project/openstack-heat/#history doesn't have 11.0.1 either19:31
smcginnisIt just switched to that name for stein.19:31
smcginnisLooks like the rc2.dev52 one listed there was probably them getting pypi set up for it.19:32
*** ianychoi has quit IRC19:32
fungihopefully it was 12.0.0-0rc1 not 12.0.0-rc119:32
fungier, not 12.0.0-rc119:32
*** ianychoi has joined #openstack-release19:33
prometheanfireya, I mistyped the dashes19:33
fungifor pyeclib, we probably need some mechanism to avoid pushing wheels to pypi since it's not pure python19:35
smcginnisWouldn't that mechanism be not having publish-to-pypi set for that repo?19:36
fungiwe could come up with infrastructure to build "manylinux1" platform binary wheels if we want, but that'll take some effort19:36
fungiwell, we can publish sdists for it to pypi, just not wheels19:36
smcginnistdasilva: ^19:37
smcginnisSo probably need a publish-to-pypi-no-wheel or something?19:37
fungipypi expressly disallows things that look like binary wheels for linux, unless they're generic manylinux wheels built to handle linking against a known minimal set of c libraries common to most distros19:38
tdasilvafungi: i guess i need to understand what changed between 1.5.0 and 1.6.0 that's breaking it being pushed to pypi...i don't think we have added non python code ??19:39
fungitdasilva: ahh, i had assumed you built c extensions against eclib or something19:39
tdasilvafungi: we do, but it's always been there19:39
fungiwas it possibly incorrectly marked as a "universal" wheel before?19:40
fungi(that's pypiese for platform-independent)19:40
fungitdasilva: hrm, last release to pypi was june of 201719:41
fungiso likely what changed in the intervening time was pypi itself19:41
smcginnispypi itself has made a lot of changes since then.19:41
fungiso have we, i think19:42
fungiat that time, we did not upload a wheel for pyeclib, only an sdist19:42
fungilikely we still had separate sdist-only jobs back then19:42
smcginnisSo is it a new job? Or some flag that can be set to skip the wheel with the current one?19:42
smcginnisI do seem to remember wheels being added fairly recently.19:43
fungigood question, not sure what to suggest, it will need some discussion when i'm not in a meeting19:43
tdasilvafungi, smcginnis: not sure yet what we can change on our side, where can I git around about skipping building a wheel?19:44
fungitdasilva: as smcginnis mentions, we probably need to modify or add to the existing release jobs, possibly even from zuul's standard library if they also similarly assume pure python packages19:45
tdasilvafungi: gotcha.19:46
fungiwe've got 22 months of bitrot for handling releases of the one non-pure-python package in openstack, it looks like19:47
tdasilva:/ sorry, that package has been widely used in swift deployments, but quite stable19:48
tdasilvafungi, smcginnis: i haven't read the scrollback (just jumped in in here due to the swift mention) so I'm not sure if this discussion came in a meeting about stein release, but FWIW, we were not planning this 1.6.0 for Stein19:51
fungitdasilva: not your fault, i think we tend to just forget openstack's python packages aren't 100% pure python19:53
smcginnistdasilva: OK, thanks. prometheanfire just brought it to our attention.19:53
*** markmcclain has quit IRC20:01
prometheanfiretdasilva: fungi: my complaint is more about tarball publishing, but pypi maters too, I guess :P20:02
fungiprometheanfire: yeah, i think the job currently tries to upload to pypi before uploading to tarballs.o.o20:03
*** pcaruana has quit IRC20:30
*** whoami-rajat has quit IRC20:30
* timburke just got back from lunch21:03
*** priteau has joined #openstack-release21:04
timburkeso i'm guessing i ought to fix .zuul.yaml for pyeclib to *just* publish an sdist (which, looking at the history on pypi, seems to be the only way we've done that previously), then tag a 1.6.1?21:04
smcginnistimburke: I think the issue is now the only publish-to-pypi job tries to build wheels by default.21:05
smcginnistimburke: So may require some work in openstack-zuul-jobs and not just in the local .zuul.yaml.21:05
smcginnisOr removal of the job I suppose and manual upload to pypi.21:06
fungithat might not be the worst workaround if it's in service of a single package which releases once every two years21:19
fungiin the spirit of https://xkcd.com/1205/21:19
fungia robust solution would probably be to have the job spot wheel platforms which pypi doesn't support and then skip uploading those to it automatically21:22
fungiwhich could in theory be handled with no changes to the interface, merely additional in-job logic21:23
fungitimburke: did you write 64945521:32
fungiafter i mentioned the above?21:32
fungior did you just read my mind? ;)21:32
fungibecause... that was really quick!21:33
timburkeyeah -- figured it'd be easier than (1) learning how to push to pypi and (2) making sure i'm allowed to push to pypi :-)21:33
timburkei'd started looking into the jobs yesterday, so i had some idea of where to look21:33
fungiwell, thanks!21:33
fungitimburke: tdasilva: smcginnis: once the zuul team merges 649455 i should be able to reenqueue the existing 1.6.0 tag for pyeclib and we'll get proper uploads21:49
timburkethanks fungi!21:49
fungii went ahead and approved it just now. we can entertain options for building out a more thorough filter there in the future if it's warranted but odds are openstack will only be building its release artifacts on 64-bit x86 linux systems for the foreseeable future anyway so that ought to be plenty sufficient for us21:52
fungiother users of that upload-pypi role outside openstack can always help improve it too21:53
smcginnisThanks fungi!21:56
timburkethe more i think about it, the more i wonder whether it should just ignore wheel upload results, and succeed/fail based on tar.gz upload21:58
timburkeseems weird to have to duplicate warehouse's whitelist in ansible; the next time they add a new supported tag, we'd have to change too21:59
fungiyeah, it's a weird scenario because twine fully expects you to just pass it a list of files you want uploaded, so the fact that we're doing them separately is already kinda weird22:14
*** priteau has quit IRC22:15
fungithough also, warehouse for the longest time (and may still) would only populate the json metadata for an entry of a wheel was the first artifact uploaded, which is why we try to upload the wheel before the sdist22:15
fungiand then there's the pragmatic bit... by failing the job outright when we're unable to upload a given wheel, we make it more obvious if there's something causing us to build a malformed wheel pypi rejects whereas if we just ignored all wheel upload failures we might not know about missing ones (and might also run afoul of things like intermittent network problems or service disruptions which caused22:16
fungione to be omitted accidentally)22:17
fungireenqueued now via `sudo zuul enqueue-ref --tenant=openstack --trigger=gerrit --pipeline=release --project=openstack/pyeclib --ref=refs/tags/1.6.0 --newrev=54eaab52d4558a630da3aa32ab5a8439f9e56ce1`22:25
fungitimburke: tdasilva: smcginnis: success! https://pypi.org/project/pyeclib/22:30
timburkewhooo! thanks guys22:33
fungiglad to help, thanks for improving the job!22:39
*** gmann_afk is now known as gmann23:11
*** tosky has quit IRC23:24
*** mlavalle has quit IRC23:33

Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!