*** threestrands has joined #zuul | 00:05 | |
*** threestrands has quit IRC | 00:05 | |
*** threestrands has joined #zuul | 00:05 | |
*** EmilienM is now known as EmilienM_PTO | 00:18 | |
*** smyers has quit IRC | 00:49 | |
*** harlowja has quit IRC | 01:05 | |
*** smyers has joined #zuul | 01:10 | |
*** swest has quit IRC | 01:56 | |
*** swest has joined #zuul | 02:11 | |
*** GonZo2000 has quit IRC | 02:13 | |
*** rlandy|rvr|bbl is now known as rlandy|rover | 02:20 | |
*** eikke has joined #zuul | 02:28 | |
eikke | Are Zuul PyPI releases supposed not to have the static web content contained in the packages? | 02:28 |
---|---|---|
*** pabelanger has left #zuul | 02:48 | |
*** myoung is now known as myoung|off | 02:57 | |
*** threestrands has quit IRC | 03:10 | |
*** myoung|off is now known as myoung|zzz | 03:12 | |
*** threestrands has joined #zuul | 03:13 | |
*** threestrands has quit IRC | 03:13 | |
*** threestrands has joined #zuul | 03:14 | |
*** threestrands has quit IRC | 03:15 | |
*** threestrands has joined #zuul | 03:15 | |
*** threestrands has quit IRC | 03:15 | |
*** threestrands has joined #zuul | 03:15 | |
*** Guest60589 is now known as trishnag | 03:35 | |
fungi | eikke: the sdist tarballs on pypi have a web subdir with static files | 04:05 |
fungi | the wheels likely don't | 04:05 |
fungi | confirmed, only in the sdist, not wheel (since wheel files are just the python bits) | 04:09 |
*** harlowja has joined #zuul | 04:47 | |
*** eumel8 has joined #zuul | 05:11 | |
*** harlowja has quit IRC | 05:11 | |
*** quiquell|off is now known as quiquell | 05:37 | |
*** GonZo2000 has joined #zuul | 05:58 | |
*** hashar has joined #zuul | 05:58 | |
*** eumel8 has quit IRC | 06:30 | |
*** GonZo2000 has quit IRC | 06:35 | |
*** threestrands has quit IRC | 07:18 | |
*** gtema has joined #zuul | 07:21 | |
*** rlandy|rover has quit IRC | 07:22 | |
eikke | fungi: I don't see those files in the sdist either? | 07:23 |
eikke | Should be in zuul/web/static | 07:23 |
eikke | they are in an sdist I create locally, because Yarn is installed. But, when Yarn is not installed, the code happily lets an sdist to be built without those artifacts | 07:24 |
*** ssbarnea_ has joined #zuul | 07:50 | |
*** jpena|off is now known as jpena | 07:53 | |
ssbarnea | am i the only one that got a "ImportError: No module named pathlib" while trying to install zuul (3.0.3) from pip? | 08:50 |
gtema | no, I have seen that yesterday trying to update my zuul in the venv. And the reason was that I was using sudo, which namely tried to install it into wrong location | 08:51 |
gtema | so source venv && pip install zuul --user solved the issue for me | 08:52 |
gtema | ok, now when I try it again locally I have it again | 08:53 |
ssbarnea | added as https://storyboard.openstack.org/#!/story/2002104 | 08:53 |
*** pcaruana has joined #zuul | 08:54 | |
gtema | ah, found. With https://github.com/openstack-infra/zuul/commit/325829b69c2e63d939e7a87669348066371407db#diff-b4ef698db8ca845e5845c4618278f29a you see that yarl 1.1.1 is used for python<3.5.3. This version seems to have broken requirements (missing pythlib) | 08:57 |
*** electrofelix has joined #zuul | 08:57 | |
gtema | so if you use python 3.6 - no problem, otherwise manually install pathlib | 08:57 |
*** pcaruana is now known as pcaruana|call| | 08:58 | |
odyssey4me | Shrews ok, so the trouble we got into was that the provider was not at quota, but it could not fulfill the request because doing so would hit quota - but it had already grabbed the request, so that request stayed in that provider's pool... and more and more of them did that. So I had one provider hogging all the requests, but unable to fulfill them until there was quota available again. Meanwhile, I had another provider with | 08:58 |
odyssey4me | quota available and nodes at a ready state doing nothing. | 08:58 |
openstackgerrit | Sorin Sbarnea proposed openstack-infra/zuul master: Add missing pathlib dependency https://review.openstack.org/570539 | 09:02 |
*** Rohaan has joined #zuul | 09:03 | |
openstackgerrit | Sorin Sbarnea proposed openstack-infra/zuul master: Add missing pathlib dependency https://review.openstack.org/570539 | 09:06 |
ssbarnea | also related to https://review.openstack.org/570546 (my attempt to use reproduce, did not went very far, two critical bugs before it really started :)) | 09:24 |
*** toabctl has joined #zuul | 10:06 | |
ssbarnea | what happened with zuul-cloner? it seems that is no longer installed by pip install zuul. | 10:07 |
gtema | ssbarnea: https://github.com/openstack-infra/zuul/commit/fc120688f85c8a98a8c0322fd5cd1796fb9e4914 | 10:09 |
ssbarnea | gtema: i found zuul cloner used inside reproduce.sh, which means that the devstack-gate which uses it no longer works. | 10:10 |
ssbarnea | see https://git.openstack.org/cgit/openstack-infra/devstack-gate/tree/functions.sh | 10:11 |
ssbarnea | there is code there calling zuul-cloner | 10:11 |
gtema | it might be, that devstack was not updated, but zuul-cloner is removed from zuul itself | 10:11 |
ssbarnea | gtema: any hints on what it would be way to fix it? | 10:12 |
gtema | ssbarnea: nope, no idea. | 10:14 |
gtema | I guess simply remove any reference | 10:14 |
gtema | most likely even, since jobs are now themselves responsible for what cloner was doing | 10:15 |
gtema | aah, see what you mean | 10:15 |
gtema | `zuul-cloner -m clonemap.yaml ...` | 10:16 |
gtema | since it is more or less just fetch me a repo - simple git clone should be sufficient | 10:21 |
gtema | there is anyway direct `git` usage later in the script | 10:22 |
*** persia has quit IRC | 10:32 | |
*** persia has joined #zuul | 10:33 | |
fungi | eikke: the sdist i retrieved from pypi and manually untarred contains zuul-3.0.3/web/util.js and so on. are those not the static web files you're looking for? | 11:21 |
fungi | gtema: ssbarnea: zuul v3 can be told to prepare a list of required projects for a job, which basically does the thing zuul-cloner used to do (remember zuul-cloner didn't just clone a got repo, it fetched and checked out specific refs zuul had prepared for testing). if you simply git clone then you'll break all your cross-repo dependencies and dependent pipelines | 11:23 |
fungi | devstack-gate has been deprecated in favor of the ansible roles which are now included in the devstack repository | 11:23 |
fungi | (that last point is technically a topic for #openstack-infra) | 11:34 |
*** jpena is now known as jpena|lunch | 11:37 | |
*** D3VIATION has joined #zuul | 11:56 | |
*** jpena|lunch is now known as jpena | 12:03 | |
*** rlandy has joined #zuul | 12:39 | |
*** rlandy is now known as rlandy|rover | 12:39 | |
*** myoung|zzz is now known as myoung | 12:42 | |
*** D3VIATION has quit IRC | 12:42 | |
ssbarnea | fungi: thanks. a clear deprecation message would have saved me some time. https://git.openstack.org/cgit/openstack-infra/devstack-gate/tree/README.rst contains no hints. | 13:28 |
*** eumel8 has joined #zuul | 13:47 | |
*** acozine1 has joined #zuul | 14:01 | |
*** quiquell is now known as quiquell|off | 14:14 | |
*** Rohaan has quit IRC | 14:16 | |
*** electrofelix has quit IRC | 14:34 | |
*** eumel8 has quit IRC | 14:55 | |
*** eumel8 has joined #zuul | 15:00 | |
*** eumel8 has quit IRC | 15:00 | |
*** bhavik1 has joined #zuul | 15:13 | |
*** bhavik1 has quit IRC | 15:18 | |
Shrews | corvus: the static driver has an option called max-parallel-jobs (number of jobs that can run in parallel on a node). That can't possibly work, can it? Doesn't zuul run jobs in the 'work' dir of the ansible user dir? This seems like parallel jobs would conflict with each other | 15:19 |
gtema | I was also wondering about that. But isn't bubblewrap for that (naive question)? | 15:23 |
tobiash | Shrews: that depends entirely on how the base job is structured | 15:23 |
tobiash | you just need a base job that can handle it | 15:23 |
tobiash | gtema: bubblewrap is used in the zuul-executor side | 15:24 |
Shrews | tobiash: that's a pretty big assumption on the part of a driver | 15:24 |
Shrews | a driver should not be dependent on a base job, IMO | 15:25 |
tobiash | Shrews: it's the other way round, the base job currently assumes that it's alone on a node | 15:25 |
tobiash | so I think the base job should be able to handle it | 15:26 |
tobiash | the driver doesn't assume anything, it just gives you several slots on a static node | 15:26 |
Shrews | but you cannot use that feature of the driver UNLESS the base job is setup to handle it | 15:26 |
Shrews | so it is a feature dependent on the base job | 15:26 |
tobiash | I think it's a limitation of the base job we should fix | 15:27 |
*** pcaruana|call| has quit IRC | 15:27 | |
gtema | tobiash: I think even our current base job is not very well able to handle this, isn't it? How would two change on the same repo tested in parallel, if the sources are packed in the ~zuul/src/git | 15:28 |
Shrews | we can't fix everyone's base jobs | 15:29 |
tobiash | Shrews: in many cases static nodes are special and often need their own base jobs anyway so we could stick a big fat warning into the docs that state that when using that feature the job needs to be able to handle concurrent runs | 15:32 |
gtema | tobiash: big fat warning is definitely what is required. In case of my zuul installation I didn't need any special base job and assumed it somehow magically work (luckily no conflicts so far) | 15:34 |
tobiash | gtema: most of the roles used by the base job use the zuul_workspace_dir (or similar) variable | 15:35 |
tobiash | so it might almost be sufficient to encode the job uuid in there | 15:36 |
tobiash | so the source is then located in ~zuul/<uuid>/src/git | 15:36 |
gtema | ok, make sence | 15:37 |
openstackgerrit | Tobias Henkel proposed openstack-infra/zuul master: Support databases on other hosts during tests https://review.openstack.org/560127 | 16:30 |
*** jpena is now known as jpena|off | 16:30 | |
*** acozine1 has quit IRC | 16:44 | |
*** rlandy|rover is now known as rlandy|rover|brb | 17:02 | |
*** gtema has quit IRC | 17:23 | |
*** gtema has joined #zuul | 17:31 | |
*** rlandy|rover|brb is now known as rlandy|rover | 17:34 | |
*** pcaruana has joined #zuul | 17:35 | |
*** myoung is now known as myoung||bbl | 17:45 | |
*** pcaruana has quit IRC | 18:14 | |
*** acozine1 has joined #zuul | 18:42 | |
*** D3VIATION has joined #zuul | 18:57 | |
*** acozine1 has quit IRC | 19:05 | |
*** myoung||bbl is now known as myoung | 19:09 | |
eikke | fungi: no. Those are the source files for the actual static HTML etc (e.g. tenants.html), which zuul-web expects to be in site-packages/zuul/web/static etc | 19:28 |
*** gtema has quit IRC | 19:45 | |
*** myoung is now known as myoung|bbl | 19:57 | |
*** harlowja has joined #zuul | 20:16 | |
*** yolanda has quit IRC | 20:34 | |
*** GonZo2000 has joined #zuul | 20:42 | |
*** GonZo2000 has joined #zuul | 20:42 | |
*** GonZo2000 has quit IRC | 20:48 | |
*** GonZo2000 has joined #zuul | 20:53 | |
*** GonZo2000 has quit IRC | 20:58 | |
*** GonZo2000 has joined #zuul | 21:10 | |
*** GonZo2000_ has joined #zuul | 21:25 | |
*** GonZo2000 has quit IRC | 21:28 | |
*** harlowja has quit IRC | 22:02 | |
tobiash | mordred: I think this is a question for you ^ | 22:04 |
tobiash | should the wheel contain the built html or is it expected that this runs during pip install? | 22:05 |
*** D3VIATION has quit IRC | 22:08 | |
*** GonZo2000_ has quit IRC | 22:23 | |
tobiash | mordred, corvus: just noticed a test failure because of a connection issue to npmjs.org: http://paste.openstack.org/show/722001/ | 22:33 |
tobiash | do we need to mirror this? | 22:33 |
*** GonZo2000_ has joined #zuul | 22:44 | |
*** hashar has quit IRC | 22:45 | |
*** myoung|bbl is now known as myoung | 22:56 | |
*** eumel8 has joined #zuul | 23:49 | |
*** rlandy|rover has quit IRC | 23:58 |
Generated by irclog2html.py 2.15.3 by Marius Gedminas - find it at mg.pov.lt!