[aodh][ec2-api][sahara][trove][vitrage][watcher] autocommit was (again) removed from LegacyEngineFacade
Hello,
Please be aware that LegacyEngineFacade in the latest oslo db release for caracal no longer enables autocommit[1]. Removal was attempted in the previous release, as part of the overall efforts to adapt to sqlalchemy 2.0, but later reverted[2] because multiple projects need more time to adapt to the change. Now the removal is again proposed for caracal and the version currently listed in upper constraints no longer provides that functionality.
Quickly searching the current usage by hound, it seems the project listed in this mail title still contains autocommit=True in LegacyEngineFacade, and is likely broken by the change. I'd suggest checking CI status to evaluate the impact.
Note 1: The list may be incomplete Note 2: It seems aodh is not actually relying on autocommit and the removal might not be affecting it
[1] https://review.opendev.org/c/openstack/oslo.db/+/875986 [2] https://review.opendev.org/c/openstack/oslo.db/+/895631
Thank you, Takashi
Hey,
Thanks for the update. I will try to look at what can be done to vitrage during the weekends. But if there's already a known way forward with that - any guidance/examples are pretty much appreciated.
ср, 6 мар. 2024 г. в 05:59, Takashi Kajinami kajinamit@oss.nttdata.com:
Hello,
Please be aware that LegacyEngineFacade in the latest oslo db release for caracal no longer enables autocommit[1]. Removal was attempted in the previous release, as part of the overall efforts to adapt to sqlalchemy 2.0, but later reverted[2] because multiple projects need more time to adapt to the change. Now the removal is again proposed for caracal and the version currently listed in upper constraints no longer provides that functionality.
Quickly searching the current usage by hound, it seems the project listed in this mail title still contains autocommit=True in LegacyEngineFacade, and is likely broken by the change. I'd suggest checking CI status to evaluate the impact.
Note 1: The list may be incomplete Note 2: It seems aodh is not actually relying on autocommit and the removal might not be affecting it
[1] https://review.opendev.org/c/openstack/oslo.db/+/875986 [2] https://review.opendev.org/c/openstack/oslo.db/+/895631
Thank you, Takashi
-- Takashi Kajinami irc: tkajinam github: https://github.com/kajinamit launchpad: https://launchpad.net/~kajinamit
Hey,
So potentially I won't refuse a hand here from somebody who has slightly more experience with sqlalchemy than me...
I was checking on work made for AODH [1] and found Vitrage code really being quite similar, except some nuances which seem to break on me. Eventually, I came with a WIP (I see quite some possible cleanups there) [2] which fails dramatically in a way I do not understand with "(sqlite3.OperationalError) database is locked".
I've tried to debug with epdb, by inserting it to the TemplatesConnection.create - if I add it inside _session_for_write() - I am able to successfully commit, close connection, rollback, whatever. But somehow, things fail on closure of _session_for_write on their own.
So help here is appreciated, as I feel I made some very stupid and obvious mistake, but I really don't understand it, since "manual" commit just works...
[1] https://review.opendev.org/c/openstack/aodh/+/896091 [2] https://review.opendev.org/c/openstack/vitrage/+/912519
ср, 6 мар. 2024 г. в 13:39, Dmitriy Rabotyagov noonedeadpunk@gmail.com:
Hey,
Thanks for the update. I will try to look at what can be done to vitrage during the weekends. But if there's already a known way forward with that - any guidance/examples are pretty much appreciated.
ср, 6 мар. 2024 г. в 05:59, Takashi Kajinami kajinamit@oss.nttdata.com:
Hello,
Please be aware that LegacyEngineFacade in the latest oslo db release for caracal no longer enables autocommit[1]. Removal was attempted in the previous release, as part of the overall efforts to adapt to sqlalchemy 2.0, but later reverted[2] because multiple projects need more time to adapt to the change. Now the removal is again proposed for caracal and the version currently listed in upper constraints no longer provides that functionality.
Quickly searching the current usage by hound, it seems the project listed in this mail title still contains autocommit=True in LegacyEngineFacade, and is likely broken by the change. I'd suggest checking CI status to evaluate the impact.
Note 1: The list may be incomplete Note 2: It seems aodh is not actually relying on autocommit and the removal might not be affecting it
[1] https://review.opendev.org/c/openstack/oslo.db/+/875986 [2] https://review.opendev.org/c/openstack/oslo.db/+/895631
Thank you, Takashi
-- Takashi Kajinami irc: tkajinam github: https://github.com/kajinamit launchpad: https://launchpad.net/~kajinamit
Ok, I think I have figured out what was blocking queries:)
вт, 12 мар. 2024 г. в 12:27, Dmitriy Rabotyagov noonedeadpunk@gmail.com:
Hey,
So potentially I won't refuse a hand here from somebody who has slightly more experience with sqlalchemy than me...
I was checking on work made for AODH [1] and found Vitrage code really being quite similar, except some nuances which seem to break on me. Eventually, I came with a WIP (I see quite some possible cleanups there) [2] which fails dramatically in a way I do not understand with "(sqlite3.OperationalError) database is locked".
I've tried to debug with epdb, by inserting it to the TemplatesConnection.create - if I add it inside _session_for_write() - I am able to successfully commit, close connection, rollback, whatever. But somehow, things fail on closure of _session_for_write on their own.
So help here is appreciated, as I feel I made some very stupid and obvious mistake, but I really don't understand it, since "manual" commit just works...
[1] https://review.opendev.org/c/openstack/aodh/+/896091 [2] https://review.opendev.org/c/openstack/vitrage/+/912519
ср, 6 мар. 2024 г. в 13:39, Dmitriy Rabotyagov noonedeadpunk@gmail.com:
Hey,
Thanks for the update. I will try to look at what can be done to vitrage during the weekends. But if there's already a known way forward with that - any guidance/examples are pretty much appreciated.
ср, 6 мар. 2024 г. в 05:59, Takashi Kajinami kajinamit@oss.nttdata.com:
Hello,
Please be aware that LegacyEngineFacade in the latest oslo db release for caracal no longer enables autocommit[1]. Removal was attempted in the previous release, as part of the overall efforts to adapt to sqlalchemy 2.0, but later reverted[2] because multiple projects need more time to adapt to the change. Now the removal is again proposed for caracal and the version currently listed in upper constraints no longer provides that functionality.
Quickly searching the current usage by hound, it seems the project listed in this mail title still contains autocommit=True in LegacyEngineFacade, and is likely broken by the change. I'd suggest checking CI status to evaluate the impact.
Note 1: The list may be incomplete Note 2: It seems aodh is not actually relying on autocommit and the removal might not be affecting it
[1] https://review.opendev.org/c/openstack/oslo.db/+/875986 [2] https://review.opendev.org/c/openstack/oslo.db/+/895631
Thank you, Takashi
-- Takashi Kajinami irc: tkajinam github: https://github.com/kajinamit launchpad: https://launchpad.net/~kajinamit
On Wed, 2024-03-06 at 13:51 +0900, Takashi Kajinami wrote:
Hello,
Please be aware that LegacyEngineFacade in the latest oslo db release for caracal no longer enables autocommit[1]. Removal was attempted in the previous release, as part of the overall efforts to adapt to sqlalchemy 2.0, but later reverted[2] because multiple projects need more time to adapt to the change. Now the removal is again proposed for caracal and the version currently listed in upper constraints no longer provides that functionality.
Quickly searching the current usage by hound, it seems the project listed in this mail title still contains autocommit=True in LegacyEngineFacade, and is likely broken by the change. I'd suggest checking CI status to evaluate the impact.
Note 1: The list may be incomplete Note 2: It seems aodh is not actually relying on autocommit and the removal might not be affecting it
[1] https://review.opendev.org/c/openstack/oslo.db/+/875986 [2] https://review.opendev.org/c/openstack/oslo.db/+/895631
Thank you, Takashi
Looks like the observability folks have merged my Aodh fix \o/ The only other outstanding project that I'm aware of is Manila. carloss and gouthamr are working on the fix for this, to the best of my knowledge.
Cheers, Stephen
participants (3)
-
Dmitriy Rabotyagov
-
Stephen Finucane
-
Takashi Kajinami