Friday, 2026-08-07

rm_work[m]tkajinam: deployed the tooz-based http-rpc chain in my staging environment this week and it has been performing great! it looks like the base tooz change just needs another +2? https://review.opendev.org/c/openstack/tooz/+/996960/02:15
rm_work[m]and I am looking into your comments on the Consul side now, need to consult with Furukawa-san I think02:15
rm_work[m]we use the Redis backend because our scale is much smaller02:15
opendevreviewAdam Harwell proposed openstack/tooz master: Add Consul coordination driver using native services  https://review.opendev.org/c/openstack/tooz/+/99714502:53
opendevreviewAdam Harwell proposed openstack/tooz master: Add k8s Lease-based coordination driver  https://review.opendev.org/c/openstack/tooz/+/99696102:53
rm_work[m]^^ k8s one is just a rebase, and TBH I am not using it anymore, considering abandoning :/02:54
opendevreviewAdam Harwell proposed openstack/oslo.messaging master: HTTP-RPC driver: tooz service discovery and fanout broadcaster  https://review.opendev.org/c/openstack/oslo.messaging/+/99747102:55
opendevreviewAdam Harwell proposed openstack/oslo.messaging master: HTTP-RPC driver: DevStack plugin, CI jobs, and documentation  https://review.opendev.org/c/openstack/oslo.messaging/+/99747202:55
stephenfinrm_work[m]: cardoe: Given I don't see either of you on #gophercloud k8s slack: I see you have different websso approachs. Would you be able to decide between yourselves which approach you'd like to do with so I can focus on one?11:45
cardoeI’m in there.11:45
cardoeBut yes we can11:45
stephenfinOh, I was foolishly searching for @cardoe šŸ˜…11:46
stephenfinrm_work[m]: Also, I saw your message RE: oslo.wsgi and ASGI but was out on paternity leave until this week. I started trying to pull out the common bits from the likes of nova, manila, and cinder (which all use broadly the same "stack") a while back but never completed it (I'd probably throw an LLM at the issue nowadays). In any case, I don't have much for you to work with yet, I'm afraid11:47
stephenfinI also had an earlier experiment to move the nova metadata service over to ASGI (using fastapi iirc) but I didn't end up seeing enough/any perf benefit to justify the churn. I saw an abandoned patch from you...somewhere...where I think you arrived at the same conclusion?11:49
cardoeI mean FastAPI just to get some of the better schema checking and error messages would be a win.12:50
-opendevstatus- NOTICE: The Gerrit service on review.opendev.openstack.org will be offline momentarily at 18:00 UTC for a patch upgrade (just under an hour from now), but should return within a few minutes.17:11
-opendevstatus- NOTICE: The Gerrit service on review.opendev.org is going offline momentarily for a patch upgrade, but should return within a few minutes.18:03
rm_work[m]stephenfin: yes I did a ton of work to get some stuff FULLY native ASGI and ... it wasn't better18:13
rm_work[m]we get the most benefit just from the frontend for http2 muxing and TLS offload18:13
rm_work[m]which I managed to get working with just an oslo factory! it works great for everything! very simple18:13
rm_work[m]I was going to pursue it further but I need to focus on the http-rpc patches first18:14
rm_work[m]then I will pick it back up18:14
rm_work[m]stephenfin: so I just did a rework of that to get it more upstreamable but, the basics of my patch have been running in our prod cloud for months and work great18:14
rm_work[m]I tried originally just using cardoe's patch but it is incomplete / didn't work for me18:15
rm_work[m]I would love to discuss though18:15
rm_work[m]I am strictly utilitarian -- if it works I do not really care which way it's implemented18:15
cardoeChange my stuff18:15
cardoeI wanna land keystoneauth-websso into keystoneauth proper as well.18:15
cardoemy stuff was I think a claude yolo.18:15
rm_work[m]mine is too but much more thorough XD18:16
rm_work[m]I think I just spent a bunch more time on revisions18:16
rm_work[m]I don't think i can join an external slack at work :(18:17
rm_work[m]ahahaha man my agent was busy last night, I told it to push up the work we did to fork branches and it decided it needed to go ahead and file additional feature requests and link everything up to the issues18:18
rm_work[m]but looks like it did it badly18:19
rm_work[m]i did not tell it to do that, but this is my first time using Codex, it seems less prompt adherent than Claude/opencode and has no plan mode that I see18:20
rm_work[m]ok, fixed them18:22
rm_work[m]TBH if it hadn't messed up the formatting that all looks... accurate18:23
rm_work[m]I did make a FEW changes to make it more upstreamable, so I will be testing those in our staging env over the next week18:58
cardoerm_work[m]: I'm game for anyway we can get websso supported in gophercloud19:50
cardoeSo closing mine and replacing it is fine with me19:50
rm_work[m]kk I will let you know how testing goes20:04
opendevreviewDaniel Bengtsson proposed openstack/oslo.service master: Ensure spawn pool workers have logging configured  https://review.opendev.org/c/openstack/oslo.service/+/98269120:23
opendevreviewDaniel Bengtsson proposed openstack/oslo.service master: Ensure spawn pool workers have logging configured  https://review.opendev.org/c/openstack/oslo.service/+/98269120:25
rm_work[m]cardoe: ok wow TBH I didn't expect new code to work perfectly first time, but it seems to :D20:31
rm_work[m]at least with the oidc part, I implement both, my CI only tests the oidc part not the websso part20:31
rm_work[m]if you can test the websso part from my branches...20:31
rm_work[m]needs all three of these together:... (full message at <https://matrix.org/oftc/media/v1/media/download/AVzIIaF9JC4HFL4dEHmsqiAz_FmYAvHDL1HadE1uNtTGqxPOnlvprRSCV2dP4RGTDWrI3suMnxQozL3qVI5bNWNCegJgYWmgAG1hdHJpeC5vcmcvbG9HRVlSbkJKU0N3cGZHRkRMUFZwSGlS>)20:32
cardoelink?20:32
rm_work[m]heh20:32
rm_work[m]I do it with go mod replace20:32
rm_work[m]I clone everything and then:... (full message at <https://matrix.org/oftc/media/v1/media/download/AVMab-X4aLofM0CQl6qkPqjaOE1Wd2d1jM3j3vFPFsq8YVm3nKqejwSWQItIqviJ0Czha0DWadrpGJlmRONY_r9CegJgfTdgAG1hdHJpeC5vcmcvVWJjVHlEblhsbW5seVFadUVLSU9Hb1Jt>)20:34
rm_work[m]oh right sorry you aren't worried about garm20:34
rm_work[m]ignore that repo :P20:35
rm_work[m]GARM is why I am doing all this20:35

Generated by irclog2html.py 4.1.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!