| opendevreview | Merged openstack/tempest master: Fix "pause_teardown" help text for docutils https://review.opendev.org/c/openstack/tempest/+/934194 | 00:33 |
|---|---|---|
| gmann | kopecmartin: ohk | 02:30 |
| *** mhen_ is now known as mhen | 02:39 | |
| opendevreview | Abhishek Kekane proposed openstack/devstack master: DNM glance s3 setup https://review.opendev.org/c/openstack/devstack/+/934307 | 08:27 |
| opendevreview | Abhishek Kekane proposed openstack/devstack master: DNM glance s3 setup https://review.opendev.org/c/openstack/devstack/+/934307 | 10:13 |
| opendevreview | Abhishek Kekane proposed openstack/devstack master: DNM glance s3 setup https://review.opendev.org/c/openstack/devstack/+/934307 | 10:30 |
| opendevreview | Abhishek Kekane proposed openstack/devstack master: DNM glance s3 setup https://review.opendev.org/c/openstack/devstack/+/934307 | 10:51 |
| fkr | hi | 13:31 |
| fkr | I'm having the problem that when installing devstack on a bare-metal host with ubuntu 22.04.5 I run into the same error each and every time (did try to re-install the host OS as well): | 13:33 |
| fkr | https://pastebin.com/WP0rizVj | 13:33 |
| fkr | no special local.conf (actually just the local.conf from the devstack docs) | 13:34 |
| clarkb | fkr: https://github.com/pypa/setuptools/issues/4483 is the error (you'll see that same mesasge in your paste). I don't know why devstack in the gate would work but your system does not. Perhaps your system comes with some preinstalled libraries that the test nodes do not (they are quite minimal no cloud-init for example) and that creats the library version collision | 15:48 |
| clarkb | either way you should check the versions of packaging and setuptools and determine how/why they are in conflict | 15:48 |
| fkr | interesting | 15:48 |
| opendevreview | Abhishek Kekane proposed openstack/devstack master: DNM glance s3 setup https://review.opendev.org/c/openstack/devstack/+/934307 | 16:14 |
| fkr | clarkb: thanks for the pointer. | 16:53 |
| fkr | clarkb: mine is a clean install from the Ubuntu 22.04 iso. Minimal install, just OpenSSH selected during install. Once installed the machine is updated via apt update/upgrade to be 22.04.5. | 17:11 |
| frickler | fkr: you can compare your "dpkg -l" output against one from our recent CI runs like https://72cca74c45a724e3198f-e25d1f8bae8601aed7ec89510f7333bc.ssl.cf1.rackcdn.com/934307/4/check/devstack/ab8455d/controller/logs/dpkg-l.txt , I think it is very likely that some delta there is the cause | 18:12 |
| frickler | you could also try to add "disable_service etcd3" to your local.conf to see if everything else is working | 18:13 |
| fkr | frickler: that I actually tried - to no avail, did not make a difference. | 18:40 |
| fkr | i'll do the re-install with the bare minimum and see wether that makes a difference. thanks for the pointer of the dpkg output - that is helpful. | 18:41 |
| fkr | (the 22.04 server installer has a minimum option that I'll try) | 18:42 |
| fkr | yes, it seems like using the "yes, just install the bare minimun" is the route to go. That is interesting, since it used to work with the other approach as well. thanks again clarkb for the hint. | 19:17 |
| sean-k-mooney | just an fyi i started running sphinx loally on watcher and im gettign build failures with the latest sphinx on py39 | 21:12 |
| sean-k-mooney | sphinx dropped supprot for 3.9 in the 7.x series | 21:12 |
| sean-k-mooney | the buidl works fin on sphinx <7.0.0 | 21:13 |
| sean-k-mooney | im going to quickly check nova but we may need to role back the version fo sphinx in uc if we actully intend our docs to be buildable on all supported python versions | 21:14 |
| sean-k-mooney | i belive all our docs jobs run on 3.10 | 21:15 |
| clarkb | sean-k-mooney: sphinx!=1.6.6,!=1.6.7,!=2.1.0,!=3.0.0,!=3.4.2 is the sphinx requirement | 21:18 |
| clarkb | I don't think anything needs to chagne? | 21:18 |
| clarkb | s/requirement/global requirement/ | 21:18 |
| clarkb | Sphinx===7.4.7;python_version=='3.9' and Sphinx===8.1.3;python_version>='3.10' is the constraint | 21:19 |
| sean-k-mooney | yep which should work based on https://github.com/sphinx-doc/sphinx/blob/v7.4.7/pyproject.toml#L16 | 21:19 |
| sean-k-mooney | however it does not actully work on rocky9 | 21:19 |
| sean-k-mooney | sphins throws https://paste.opendev.org/show/bcqW7NsmVEI5WC0Kp8Oj/ | 21:20 |
| clarkb | the exception originates in openstackdocstheme not sphinx | 21:21 |
| sean-k-mooney | i gues the error is actully a combination of the sphinx version, the python version and openstackdocs theme | 21:21 |
| sean-k-mooney | right but using an older version of sphinx works | 21:21 |
| clarkb | relative_to is a pathlib method | 21:21 |
| sean-k-mooney | i can try downgrading openstackdocsthemem instead | 21:21 |
| clarkb | so on newer python you get paths as pathlib objects and older python you get strings? but that problem seem to be in the docs theme | 21:22 |
| sean-k-mooney | yes | 21:22 |
| sean-k-mooney | so i think we actuly fix that in a few other places already | 21:22 |
| clarkb | https://github.com/openstack/openstackdocstheme/blob/master/openstackdocstheme/page_context.py#L83-L88 | 21:22 |
| clarkb | the code is sphinx version conditional | 21:22 |
| clarkb | maybe you aren't getting the older sphinx version? or that check is broken | 21:23 |
| sean-k-mooney | no according to pip freeze i am | 21:23 |
| sean-k-mooney | so the check is likely broken | 21:23 |
| sean-k-mooney | >>> sphinx.version_info | 21:24 |
| sean-k-mooney | (7, 4, 7, 'final', 0) | 21:24 |
| sean-k-mooney | >>> sphinx.version_info >= (7, 0) | 21:25 |
| sean-k-mooney | True | 21:25 |
| clarkb | oh I see the problem is its checkig >= 7 not 8 | 21:25 |
| sean-k-mooney | i think that should be >= (8,0) | 21:25 |
| clarkb | ya could be | 21:25 |
| sean-k-mooney | ya | 21:25 |
| sean-k-mooney | so 8 drops support for 3.9 | 21:25 |
| sean-k-mooney | which i guess is why we have the current pin to 7.4.7 on 3.9 | 21:26 |
| clarkb | https://github.com/openstack/openstackdocstheme/commit/f00d2113281aacba3b59337c68221e33a33ef7d1 | 21:26 |
| clarkb | ya I think the issue is just having the pivot point at the wrong location | 21:26 |
| clarkb | its also possible 7.0 on python3.10 gives you a pathlib and with python3.9 it gives you a string so have two axis on the check | 21:27 |
| sean-k-mooney | so there are a few ways to fix it | 21:28 |
| sean-k-mooney | we could clamp the docstheme only on 3.9 to < 3.4.0 | 21:28 |
| sean-k-mooney | ill check if that works locally | 21:28 |
| sean-k-mooney | ya that works | 21:30 |
| sean-k-mooney | so if pathlib came in 3.10 then we can clamp to openstackdocstheme==3.3.0 only on 3.9 | 21:30 |
| clarkb | pathlib has been around since 3.4 but you have to opt into using it I think? | 21:30 |
| clarkb | and sphinx maybe does that conditionally on python version for some reason (feature availability?) | 21:31 |
| sean-k-mooney | so i had to do this https://github.com/openstack/nova-specs/commit/39c91027a63c9e9551aa875359cbf94e04302d2e | 21:31 |
| clarkb | looking at that commit its addressing warnings I wuold just chagne the pivot point | 21:31 |
| clarkb | the problem with pinning the docs theme is youdon't want the docs to come out different depending on your python version | 21:32 |
| sean-k-mooney | becuase i think app.builder.srcdir started beign a pathlib object | 21:32 |
| sean-k-mooney | ok well the simplest thing is to push a patch to the repo to move the pivot | 21:32 |
| sean-k-mooney | and see if it passes ci there | 21:33 |
| sean-k-mooney | i do not see any 3.9 testing in https://github.com/openstack/openstackdocstheme | 21:33 |
| sean-k-mooney | my guess is its all runing on 3.10 by using the defualt nodeset | 21:34 |
| clarkb | ya doesn't seem to run unittests but does run docs generation. Might want a docs under different python versions set of jobs in that repo | 21:36 |
| sean-k-mooney | porably the extreams min/max supproted | 21:36 |
| clarkb | ++ | 21:36 |
| sean-k-mooney | im not going to do that tonight but ill push a patch for 8 and we can see what other think in the morning | 21:37 |
| sean-k-mooney | ok so tox -e docs fails on openstackdocstheme without the change and works with 8 | 21:41 |
| sean-k-mooney | ill add a commit message and push that, thanks for the help clarkb | 21:42 |
| clarkb | cool I think that is the best outcome to keep doc generation consistent across python versions | 21:43 |
| clarkb | would just be odd if there ended up being a difference like that (I have no idea if we would have one today but could if we split the versions) | 21:44 |
| sean-k-mooney | yep for now ill just manually pin it locally, im currently enableing pre-commit and sphinx lint but i want ot make sure i dont break the docs build in the process | 21:44 |
| sean-k-mooney | i just happend to be useing 3.9 becuase i was testing somethign else that only breaks on 3.9 | 21:44 |
| sean-k-mooney | clarkb: ill ping stephen about https://review.opendev.org/c/openstack/openstackdocstheme/+/934384 on monday. i think he is at a wedding tomorrow | 21:50 |
| sean-k-mooney | im not sure who else actully maintains that at this point | 21:51 |
| JayF | your assumption that he is it is mostly correct, I think it came up recently in #openstack-tc or a tc meeting | 22:06 |
| *** haleyb is now known as haleyb|out | 22:44 | |
| *** tosky_ is now known as tosky | 23:40 | |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!