| opendevreview | Joan Gilabert proposed openstack/watcher master: Set number of decimal digits in efficacy indicator https://review.opendev.org/c/openstack/watcher/+/945199 | 10:41 |
|---|---|---|
| sean-k-mooney | jgilaber: so to fix the installation issue we need to do this https://review.opendev.org/c/openstack/placement/+/919569 and then do this https://review.opendev.org/c/openstack/placement/+/932399 | 12:31 |
| sean-k-mooney | + adtapt devstack the devstack pluging to use the module path instead of a file path | 12:31 |
| sean-k-mooney | which looks like this https://review.opendev.org/c/openstack/devstack/+/919581/2/lib/placement | 12:32 |
| sean-k-mooney | because fo how ci is curently broken we will need to merge teh devstack plugin change first. possible making it non voting before we can merge the watcher change | 12:33 |
| sean-k-mooney | but ill see if i can do it without needing to disabel the jobs | 12:33 |
| sean-k-mooney | we also likely need ot combine this with droping py39 supprot | 12:34 |
| jgilaber | in our case we don't have any dependency to the devstack repo right? | 12:34 |
| jgilaber | we should be able to put all the fixes in one patch | 12:34 |
| jgilaber | it'll big, but it could unblock CI in one go | 12:35 |
| sean-k-mooney | we need to update the devstack plugin but ya thats in tree | 12:39 |
| sean-k-mooney | so we shoudl be able to do everythign in one patch | 12:39 |
| sean-k-mooney | we may need to backport this too in which case we may need to drop part of it but we will see how stable looks | 12:40 |
| sean-k-mooney | ill file an upstream bug for this to track it in case we need to backport | 12:41 |
| jgilaber | sounds good, thanks sean-k-mooney | 12:46 |
| sean-k-mooney | the only reason i know about this is i found this problem 18 months ago when trying to get devstack on centos to work in a venv. stephen and i have fixed this in a number of repos but not all of them | 12:48 |
| sean-k-mooney | look like anyone who has not fixed it yet will have to do it now | 12:48 |
| sean-k-mooney | pip had a new release 3 days ago so my guess is now there is no choice but to implementit | 12:49 |
| sean-k-mooney | we have been relying on deprecated compatiablity code that was added in 2023 and was ment to be removed last year | 12:50 |
| sean-k-mooney | so we cant really blame pip for this | 12:50 |
| dviroel | thanks sean-k-mooney and jgilaber | 12:59 |
| dviroel | I will review once it is ready | 12:59 |
| *** haleyb|out is now known as haleyb | 13:00 | |
| sean-k-mooney | dviroel: this is why my ruff patch was failing ci for what its worth. i added a pyproject.toml but didnt do the wsgi module and devstack change | 13:00 |
| sean-k-mooney | i just interestign timing that pip/requiremetn repo changes happend this week | 13:01 |
| sean-k-mooney | i filed https://bugs.launchpad.net/watcher/+bug/2109608 to rack it | 13:02 |
| sean-k-mooney | *track | 13:02 |
| dviroel | yeah, at least, it didn't happen in feature freeze week ^^ | 13:02 |
| dviroel | ack | 13:02 |
| sean-k-mooney | i was aware of this technial debt form teh review i did last cycle i just dint have it high on my list but ya FF week woudl have been stressful | 13:03 |
| opendevreview | sean mooney proposed openstack/watcher master: Add support for pyproject.toml and wsgi module paths https://review.opendev.org/c/openstack/watcher/+/948438 | 13:53 |
| opendevreview | sean mooney proposed openstack/watcher master: drop suse supprot in the devstack plugin https://review.opendev.org/c/openstack/watcher/+/948439 | 13:53 |
| opendevreview | sean mooney proposed openstack/watcher master: drop jammy jobs https://review.opendev.org/c/openstack/watcher/+/948440 | 13:53 |
| sean-k-mooney | so i have not had time to test those locally but i belive the first patch will fix ci. | 13:53 |
| sean-k-mooney | and the other two are just cleanup | 13:53 |
| jgilaber | sean-k-mooney: thanks, I'll give it a try in my deployment | 14:00 |
| dviroel | i spot a small issue thre | 14:08 |
| sean-k-mooney | yep | 14:39 |
| sean-k-mooney | im trying to decied how deep down the rabit whole i want to go in one patch | 14:39 |
| sean-k-mooney | i have undone some of the chagne i have made an put it back into the wsgi module | 14:39 |
| sean-k-mooney | instead of __init__.py | 14:39 |
| sean-k-mooney | i should rename the init funciton but i think ill do that in a followup | 14:40 |
| sean-k-mooney | and i have some other funcitontionality to remove too | 14:40 |
| jgilaber | sean-k-mooney: I just tried applying the small issues that dviroel pointed out and it seems to work | 14:48 |
| jgilaber | use the "api.initialize_wsgi_app" method seems fine, fyi | 14:49 |
| sean-k-mooney | yep bu t i think some wsgi server expect the other name | 14:51 |
| sean-k-mooney | im reverting to that initialize_wsgi_app for now | 14:52 |
| sean-k-mooney | and ill look at changing it in a follow up if needed | 14:52 |
| dviroel | we will need to move to another wsgi due to eventlet removal too :( | 15:18 |
| dviroel | in a near future | 15:18 |
| sean-k-mooney | kind of | 15:20 |
| sean-k-mooney | we just need to delete the console script | 15:20 |
| sean-k-mooney | but we need to deprecate it | 15:20 |
| sean-k-mooney | i dont think we are relying on the wsgi module form oslo service in teh wsgi appclation just in the watcher-api console script | 15:20 |
| opendevreview | sean mooney proposed openstack/watcher master: Add support for pyproject.toml and wsgi module paths https://review.opendev.org/c/openstack/watcher/+/948438 | 15:27 |
| opendevreview | sean mooney proposed openstack/watcher master: drop suse supprot in the devstack plugin https://review.opendev.org/c/openstack/watcher/+/948439 | 15:27 |
| opendevreview | sean mooney proposed openstack/watcher master: drop jammy jobs https://review.opendev.org/c/openstack/watcher/+/948440 | 15:27 |
| dviroel | sean-k-mooney: https://opendev.org/openstack/watcher/src/commit/c4acce91d6bb87b4ab865bc8e4d442a148dba1d5/watcher/common/service.py#L83 ? | 15:27 |
| sean-k-mooney | ya that using the eventlet webserver | 15:28 |
| sean-k-mooney | to host the wsgi applcition | 15:28 |
| sean-k-mooney | but the wsgi appcltion does not depend on eventlet or oslo service's wsgi moducle as far as im aware | 15:29 |
| sean-k-mooney | dviroel: we deelted nova's equivalte of this last week or the week before | 15:29 |
| sean-k-mooney | but the thing we need to check in watcher is have we deprecated it yet | 15:30 |
| sean-k-mooney | if we havng then in theroy we shoudl not delete it until 2026.2 | 15:30 |
| sean-k-mooney | dviroel: the actul freestandign wsgi aplication is here https://opendev.org/openstack/watcher/src/commit/c4acce91d6bb87b4ab865bc8e4d442a148dba1d5/watcher/api/wsgi.py | 15:31 |
| dviroel | yeah, the pecan app | 15:31 |
| sean-k-mooney | dviroel: where as that is sued form here https://opendev.org/openstack/watcher/src/commit/c4acce91d6bb87b4ab865bc8e4d442a148dba1d5/watcher/cmd/api.py#L39 | 15:31 |
| sean-k-mooney | the watcher-api console script that calles that cmd module's main https://opendev.org/openstack/watcher/src/commit/c4acce91d6bb87b4ab865bc8e4d442a148dba1d5/setup.cfg#L42 | 15:32 |
| sean-k-mooney | will just go away with no replacemnt | 15:32 |
| sean-k-mooney | as will https://opendev.org/openstack/watcher/src/commit/c4acce91d6bb87b4ab865bc8e4d442a148dba1d5/setup.cfg#L49-L50 eventualy but that not related to eventlet removal | 15:32 |
| sean-k-mooney | what will remain is the actual wsgi appcltion | 15:33 |
| dviroel | yeah, correct | 15:33 |
| sean-k-mooney | so unfortunetly it looks like watcher has not deprecated its eventlet watcher-api console script yet | 15:35 |
| sean-k-mooney | so we shoudl do that and then talk to the stable team about if we can/shoudl backprot that | 15:35 |
| sean-k-mooney | technicaly speakign however we are required to supprot tha tfor upgrades form 2025.1 to 2026.1 because we didnt deprecate it in 2025.1 | 15:36 |
| sean-k-mooney | dviroel: looks like i still have an erro in my updated patch but im going to deploy locally and test it in a clean devstack | 17:52 |
| sean-k-mooney | i proably wont have a new version up today but we should be able to get this fixed tomorrow | 17:52 |
| dviroel | ok, np, thanks sean-k-mooney | 17:53 |
| opendevreview | sean mooney proposed openstack/watcher master: Add support for pyproject.toml and wsgi module paths https://review.opendev.org/c/openstack/watcher/+/948438 | 19:04 |
| opendevreview | sean mooney proposed openstack/watcher master: drop suse supprot in the devstack plugin https://review.opendev.org/c/openstack/watcher/+/948439 | 19:04 |
| opendevreview | sean mooney proposed openstack/watcher master: drop jammy jobs https://review.opendev.org/c/openstack/watcher/+/948440 | 19:04 |
| opendevreview | sean mooney proposed openstack/watcher master: Add support for pyproject.toml and wsgi module paths https://review.opendev.org/c/openstack/watcher/+/948438 | 20:16 |
| opendevreview | sean mooney proposed openstack/watcher master: drop suse supprot in the devstack plugin https://review.opendev.org/c/openstack/watcher/+/948439 | 20:16 |
| opendevreview | sean mooney proposed openstack/watcher master: drop jammy jobs https://review.opendev.org/c/openstack/watcher/+/948440 | 20:16 |
Generated by irclog2html.py 2.17.3 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!