| sean-k-mooney | clarkb: fungi is https://review.opendev.org/c/openstack/project-config/+/951395 something we can now move forward with since the release is done? im asking because i belive its require to make https://review.opendev.org/c/openstack/grian-ui/+/982397 unessisary | 12:37 |
|---|---|---|
| sean-k-mooney | when i setup grian-ui i made it https://peps.python.org/pep-0639/ complaint and used the string for the licens instead of the deprecated dict form | 12:38 |
| sean-k-mooney | we need to fix long_description seperatly to make test-release-openstack happy | 12:39 |
| sean-k-mooney | but i think the fact pep-0639 format is not supproted is beacuse we are building on such an old distro | 12:40 |
| sean-k-mooney | hence why ima skignif we can proceed with the move form focal to noble | 12:40 |
| frickler | approved, we can always revert if something breaks and we're well after the release now | 12:43 |
| sean-k-mooney | thanks :) | 12:44 |
| sean-k-mooney | that fact thats in a config repo makes it harder to test with | 12:44 |
| sean-k-mooney | those jobs are not using secrets right | 12:44 |
| sean-k-mooney | woudl it ever make sense to move those to the relsases repo? | 12:45 |
| sean-k-mooney | at least the test-release-openstack definition | 12:45 |
| sean-k-mooney | publish-openstack-python-tarball is lightly idffent as that does need secret access | 12:46 |
| sean-k-mooney | there si some advantage to keepign them both in the oen repo however so im just contemplating fi there is any advantage to that | 12:47 |
| opendevreview | Merged openstack/project-config master: Update nodeset for test-release-openstack https://review.opendev.org/c/openstack/project-config/+/951395 | 12:53 |
| fungi | sean-k-mooney: the test-release-openstack job does not use any secrets, no | 13:50 |
| sean-k-mooney | for what its worth they liecense issue is not resolved and i just need to fix the readme and it should pass on grain-ui | 13:53 |
| sean-k-mooney | i have replicated some fo those chackes locally in a tox target so we can debug those failure trivially gong forward | 13:54 |
| sean-k-mooney | https://review.opendev.org/c/openstack/grian-ui/+/983723 | 13:54 |
| sean-k-mooney | althought the version may not align ot what the job is useing but its better then rechecks :) | 13:54 |
| sean-k-mooney | .. TODO:: is apprently a sphinx extsions which was news to me | 13:56 |
| sean-k-mooney | or at least its not supprote by twine | 13:56 |
| fungi | `twine --check` will test that for you locally if you want | 14:04 |
| opendevreview | Stephen Finucane proposed openstack/project-config master: Add GitHub mirroring for devstack-plugin-prometheus https://review.opendev.org/c/openstack/project-config/+/983728 | 14:17 |
| stephenfin | fungi: In the context of the above, is there any specific criteria for official projects when it comes to devstack plugins? Is it anything in the openstack namespace? | 14:41 |
| fungi | stephenfin: i think the repositories get auto-created on github based on a periodic (daily) job that runs, so if it doesn't exist there yet (rather than being merely empty/stale) then it's likely based on release or governance metadata, i don't recall which | 14:43 |
| fungi | oh, right, it's the same job that runs the pull request closer, lemme see if i can find it | 14:43 |
| fungi | this one: https://zuul.opendev.org/t/openstack/job/maintain-github-openstack-mirror | 14:44 |
| fungi | though it's been failing since early january | 14:45 |
| stephenfin | Yeah, I spotted that. It's handled by the `official-openstack-repo-jobs` template, but the naming of that suggested there would be a formal definition of what "official openstack" was | 14:45 |
| stephenfin | s/handled/included/ | 14:45 |
| fungi | this is the meat of it: https://opendev.org/openstack/project-config/src/branch/master/playbooks/maintain-github-mirror/github_manager.py | 14:46 |
| fungi | and it's based on governance metadata | 14:48 |
| fungi | reference/{projects,sigs-repos,technical-committee-repos,user-committee-repos}.yaml specifically | 14:49 |
| fungi | but also by tc policy, projects in the openstack/ namespace need to be listed in one of those files (or legacy.yaml for retired projects) | 14:49 |
| fungi | so i think the answer to your question is yes on both: all projects in the openstack/ namespace and projects listed in governance, because to be in one they're expected to also be in the other | 14:50 |
| fungi | except retired projects don't get auto-created there, for obvious reasons | 14:51 |
| *** bbobrov_ is now known as bbobrov | 15:00 | |
| JayF | Noting the googly issues have gotten worse; I'm having to search for openstack docs in bing to get any reasonable results. :( I know this isn't directly an infra thing, but there's no point in us publishing docs at all if users can't find them :| | 15:44 |
| sean-k-mooney | stephenfin: offically openstack prroject are onder the governace of the tc and a project team | 15:47 |
| sean-k-mooney | but anywone can create a devstack plugin in there own namespace | 15:48 |
| sean-k-mooney | or on github ectra | 15:48 |
| sean-k-mooney | stephenfin: was there a specific plugin you were interested ing | 15:48 |
| stephenfin | sean-k-mooney: Yes, this one https://review.opendev.org/c/openstack/project-config/+/983728 | 15:49 |
| sean-k-mooney | oh ya so we created that after the default mirror was disable | 15:50 |
| sean-k-mooney | there is no issue form my part on doing that | 15:50 |
| sean-k-mooney | this is co maintained by watcher/telemetry teams | 15:50 |
| stephenfin | yeah, as mandre notes we want the mirror for Gophercloud CI (since that runs on GitHub, sadly) | 15:50 |
| frickler | we try to keep a full list here https://docs.openstack.org/devstack/latest/plugin-registry.html | 15:51 |
| sean-k-mooney | stephenfin: as noted above however the officalnes of the proejct does not impact if its mirror or not | 15:52 |
| sean-k-mooney | its more that you have to opt in to the mirror now and we often skip that step for newly crfeated repos | 15:53 |
| stephenfin | Gotcha. Sounds like what I've done is okay so (an opinion reinforced by frickler's ack) | 15:55 |
| sean-k-mooney | ya the other option woudl have been to add the template in the openstack/devstack-plugin-prometheus | 15:56 |
| sean-k-mooney | repo | 15:56 |
| sean-k-mooney | but we conventioally do it in the project-cofnig so that is fine | 15:56 |
| stephenfin | I believe that's a "protected" job that can only be run from that repo | 15:58 |
| stephenfin | At least I don't see any references to that template in other repos | 15:58 |
| stephenfin | s/references to/users of/ | 15:58 |
| kozhukalov | https://www.irccloud.com/pastebin/iBSMcvEw/ | 16:02 |
| clarkb | kozhukalov: I think this discussion was had with fungi previously. I think the idea at the time was to create a dns record to effectively document the name (maybe a TXT or something?) but that should be doable | 16:06 |
| kozhukalov | Ah, ok, probably @cardoe have already raised this question. | 16:07 |
| kozhukalov | Will ask him for details. | 16:07 |
| cardoe | kozhukalov: yep. I asked for that already. | 16:07 |
| cardoe | fungi wanted to make a DNS entry so it wasn't forgotten | 16:08 |
| kozhukalov | Now i don't see any recoreds with this name. Should we ask him again? Or maybe there should be a formal request? | 16:10 |
| clarkb | I think we've just all been incredibly bsuy. Between openstack release, various vmt things, the web crawlers, migrating email, gerrit upgrade prep, gazpacho metric mangling, etc there is only so much the small number of people working on this can do | 16:11 |
| fungi | yes, i can add it, more a question of what i point it to. we could create a redirect site on our static hosting that redirects that name to the openstack-helm docs, for example | 16:12 |
| clarkb | fungi: I would just create a txt record "path used for OSH k8s CRDs" or whatever | 16:12 |
| clarkb | I wouldn't point it to anything imo | 16:12 |
| fungi | a normal dns lookup won't find txt records, someone would need to specifically ask for that | 16:12 |
| clarkb | but that will serve as signal to anyone who might try to add an A record that hey this name is arleady in use | 16:12 |
| clarkb | right but this won't ever be used for dns its just to prevent collisions later? | 16:13 |
| fungi | also a txt record in the domain wouldn't prevent someone from adding a/aaaa records for the same name | 16:13 |
| fungi | or necessarily even be noticed if someone tried, since dns management interfaces often sort by record types | 16:13 |
| clarkb | oh that is unfortunate. I guess the rax dns did that ya | 16:14 |
| clarkb | I don't know what cloudflare does. i would consider that a bug :) | 16:14 |
| fungi | as does cloudflare, where openstack.org is hosted, yes | 16:14 |
| clarkb | but also I don't think anyone cares about my dns record editing opinions | 16:14 |
| fungi | remote: https://review.opendev.org/c/opendev/system-config/+/983757 Add an osh.openstack.org redirect site [NEW] | 16:25 |
| fungi | cardoe: kozhukalov: clarkb: ^ something like that, then i just cname osh.openstack.org to static.opendev.org in dns | 16:25 |
| cardoe | Sounds good to me. | 16:26 |
| clarkb | fungi: I think you should not use a wildcard redirect | 16:26 |
| clarkb | fungi: maybe just / -> latest/ | 16:26 |
| fungi | good call, i duplicated how the legacy nova.openstack.org redirect is configured, but yes that's been a source of pain for us in the recent past | 16:27 |
| kozhukalov | Looks good | 16:27 |
| fungi | removed the (.*) in both vhosts | 16:27 |
| sean-k-mooney | fungi: i was about to ask if we actully have someint in dns for the xml namesace domain we use for nova or not | 16:34 |
| fungi | i don't know about any xml namespace, but if it's nova.openstack.org then yes we have a working dns redirect to the nova docs | 16:35 |
| sean-k-mooney | its not that mibt be the wone we sue for the json schema for the metadta stuff | 16:36 |
| sean-k-mooney | NOVA_NS = "http://openstack.org/xmlns/libvirt/nova/1.1" | 16:36 |
| sean-k-mooney | that what we use for the metadta namesapce in teh domain xml | 16:36 |
| clarkb | openstack.org isn't managed by us so I don't think we know | 16:36 |
| clarkb | but that also probably means no | 16:37 |
| sean-k-mooney | we dont use nova.openstack.org in the code base | 16:37 |
| sean-k-mooney | it just the docs redirect | 16:37 |
| fungi | yeah, to clarify, the http(s)://openstack.org/ site and content are not community-managed, it's outsourced by the openinfra foundation to a webdev contract firm | 16:37 |
| sean-k-mooney | nova never actully published a schema ther anyway | 16:38 |
| sean-k-mooney | it just need to be set to soemthing for xml parser to be happy | 16:38 |
| sean-k-mooney | the intent is you set it to a location where the validation schema can be retrived from if client supprot that but libvirt does not actully use it | 16:39 |
| sean-k-mooney | having osh.o.o redirect to the docs is alos nice fo discoverabliy i guess if some one does try to check it out | 16:40 |
| clarkb | theoretically you could publish the schema to docs.openstack.org/nova or whatever then use that url | 16:40 |
| sean-k-mooney | we do that for one of our schemas but we dont actuly have a schemea for this | 16:41 |
| sean-k-mooney | we just generate the xml form phython classes but we dont genearte a validation schema form those classes | 16:41 |
| *** skandix4263990 is now known as skandix426399 | 16:54 | |
| clarkb | JayF: we've got wes doing things with the google domain dashboard stuff includign loading a sitemap.xml (the one published by openstack manuals). The idea is to see if we can push google to reload what it knows of the site | 19:46 |
| clarkb | and if that helps great. If not we can try the next thing. Which may involve improving the sitemap.xml to better map out what is important? | 19:46 |
| fungi | https://docs.openstack.org/sitemap.xml specifically | 19:46 |
| fungi | and apparently, as clarkb observed (based on errors reported by google's sitemap parser), that file is incorrectly using iso 8601 formatted timezone offsets instead of w3c datetime offsets | 19:54 |
| fungi | https://www.sitemaps.org/protocol.html | 19:54 |
| clarkb | should be +00:00 instead of +0000 | 19:54 |
| fungi | i suspect it's ultimately coming from https://opendev.org/openstack/openstack-doc-tools/src/commit/c6b5d46/sitemap/generator/spiders/sitemap_file.py#L133 | 19:55 |
| fungi | er, the line right after that actually, L134 | 19:55 |
| fungi | doesn't look like strftime() has inbuilt support for breaking up the hours and minutes in the offset: https://docs.python.org/3/library/time.html#time.strftime | 19:57 |
| fungi | but some quick string munging could take care of that as a separate call | 19:58 |
| JayF | clarkb: I appreciate the follow-up :) | 19:59 |
| fungi | remote: https://review.opendev.org/c/openstack/openstack-doc-tools/+/983790 Correct lastmod time zone offset in sitemaps [NEW] | 20:05 |
Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!