Friday, 2026-01-23

*** mhen_ is now known as mhen02:46
opendevreviewIvan Anfimov proposed openstack/freezer master: wip  https://review.opendev.org/c/openstack/freezer/+/97439908:13
opendevreviewIvan Anfimov proposed openstack/freezer master: wip  https://review.opendev.org/c/openstack/freezer/+/97439908:17
opendevreviewIvan Anfimov proposed openstack/freezer master: wip  https://review.opendev.org/c/openstack/freezer/+/97439908:21
opendevreviewIvan Anfimov proposed openstack/freezer master: docs: fix few mistakes  https://review.opendev.org/c/openstack/freezer/+/97439908:22
opendevreviewIvan Anfimov proposed openstack/freezer master: docs: fix few mistakes  https://review.opendev.org/c/openstack/freezer/+/97439908:31
noonedeadpunkhuh, seems that smth is off with devstack....10:25
noonedeadpunkI'm really not sure if it's the freezer issue or just devstack in general10:27
noonedeadpunklike we haven't merged anything which would make API to fail to start I'd say10:29
noonedeadpunkhttps://zuul.opendev.org/t/openstack/build/9cb2365018b7402aa9b4d985c6a92b37/log/controller/logs/screen-freezer-api.txt\10:29
opendevreviewDmitriy Rabotyagov proposed openstack/freezer-api master: Use SSL and authentication for elasticsearch  https://review.opendev.org/c/openstack/freezer-api/+/97441111:59
mhennoonedeadpunk: yea that looks really strange, I have no explanation either12:09
mhenI compared to the logs of my running instance on my devstack and even the "UNABLE to load uWSGI plugin" at the top is there so that is not the culprit12:10
mhenotherwise it looks pretty similar up until "unable to load app 0"12:11
mhenno clue12:11
noonedeadpunkMy guess was that some dependency was updated in u-c, but I have not figured out which ones12:13
noonedeadpunk(like flask or smth)12:13
noonedeadpunkbtw, elasticsearch driver seems way more broken then anticipated. It also seems not really compatible with python library version after the upgrade12:14
noonedeadpunktrying to figure this out now12:14
noonedeadpunkmhen: I think this is the reason: https://opendev.org/openstack/requirements/commit/be5dd6e6b69ef23d6a2ba87041dbc2c39f55575013:32
noonedeadpunkhttps://opendev.org/openstack/oslo.db/commit/4a29bee924bdf1636e797151ce79a05daddd51f7 -> freezer-api relises on this still13:34
noonedeadpunkI think for elastic to work, we actually need to significantly refactor it....13:58
mhennoonedeadpunk: currently looking into making the naming of resources consistent, especially the temporary ones; what do you think about naming the final restored resources? The glance mode prepended "restore_" but cinder and nova tried to replicate the original name; I'm thinking about "original-name" vs. "original-name (restored) here.14:56
mhenThe "restore_" could be confused with the temporary resource prefix, so I wouldn't want to use that for the final resources14:57
mhennaming them the same as the original resources can get confusing if the original ones still exist14:58
noonedeadpunkI think I'd avoid using brackets or spaces potentially, but not sure14:58
noonedeadpunkbut also description can help to distinguish14:59
mhenproblem is I realized that glance images have no description field D:14:59
mhenso I had to drop that idea14:59
mhenand glance images are used a lot as temporary resources15:00
mhenI could do it for servers and volumes but I want to achieve consistency across the board here15:00
noonedeadpunkok, right15:02
noonedeadpunkI'd say that having also the backup ID from which it was restored might be useful then?15:02
mhendifficult; I think backups do not have an ID per se, only an optional name15:04
mhenI think the most reliable and meaningful thing would be the backup timestamp15:05
noonedeadpunkah... becauys API is optional, and it's different also for elastic/mysql?15:09
noonedeadpunkas I opriginally looked at https://opendev.org/openstack/freezer-api/src/branch/master/freezer_api/db/sqlalchemy/models.py#L18015:09
mhenI'm looking at freezer-agent isolated; it does not have the notion of a backup id when used directly15:10
noonedeadpunkI kind of dunno about restoration name. not that I have some opinion there tbh. 15:10
noonedeadpunkas it can be confusing both ways.15:11
noonedeadpunklike you could easily do multiple restores in attempt to look for some valid ones15:11
noonedeadpunkand having 5 restored resources with the same name would be also confusing15:11
mhenI think "<name>@<timestamp>" would be the most clear way to differentiate actual different instances of a restored resource and help identifying its origin at a glance (due to the timestamp)15:13
mhenas in timestamp of the point in time the backup was created15:15
noonedeadpunkyeah, just be super carefull with dividers I'd say15:22
noonedeadpunkso that there was no need for some kind of urlsafe things for clients/autopmation/etc15:23
mhenfair point15:31
opendevreviewDmitriy Rabotyagov proposed openstack/freezer-api master: Remove LegacyEngineFacade  https://review.opendev.org/c/openstack/freezer-api/+/97444715:58
noonedeadpunkhopefully this will fix CI ^15:59
opendevreviewDmitriy Rabotyagov proposed openstack/freezer-api master: Remove LegacyEngineFacade  https://review.opendev.org/c/openstack/freezer-api/+/97444716:05
opendevreviewMarkus Hentsch proposed openstack/freezer master: Consistent & meaningful OpenStack resource naming  https://review.opendev.org/c/openstack/freezer/+/97448317:04
-opendevstatus- NOTICE: The Gerrit service on review.opendev.org will be offline momentarily while we restart it onto new images after an update to the build toolchain18:01
*** gmaan is now known as gmaan_afk19:39

Generated by irclog2html.py 4.0.0 by Marius Gedminas - find it at https://mg.pov.lt/irclog2html/!