| Posted | Nick | Remark | |
|---|---|---|---|
| #openstack-nova - 2023-10-06 | |||
| 07:26:07 | zigo | (we're trying to upgrade from Victoria to Bobcat...) | |
| 07:37:38 | frickler | if you see doc build failures please check https://bugs.launchpad.net/nova/+bug/2026345 | |
| 08:14:38 | bauzas | zigo: with alembic, that's a bit different but you can find them | |
| 08:15:04 | zigo | Yeah, I found the version in the db, but can't find the version in the code ... | |
| 08:17:56 | auniyal | frickler, I am seeing this today, even though my change do not have any doc update, thanks for sharing bug. | |
| 08:21:01 | frickler | zigo: there was a big cleanup in https://review.opendev.org/c/openstack/nova/+/872428 which might affect you? | |
| 08:22:29 | frickler | are you trying to go in one big step from V to B? | |
| 08:37:55 | zigo | frickler: No, we're currently designing our upgrade script to script it all... | |
| 08:38:09 | zigo | And we'd like to add db version checks on each step. | |
| 08:40:13 | zigo | So we need this check for both alembic and sqla-migrate... | |
| 08:41:59 | zigo | So far, I have: | |
| 08:41:59 | zigo | SELECT version_num FROM alembic_version; | |
| 08:41:59 | zigo | to get the current version in the *db* for alembic, and: | |
| 08:41:59 | zigo | python3 manage.py version | |
| 08:42:00 | zigo | to get the *code* version of sqla-migrate. I need to find out how to fetch: | |
| 08:42:02 | zigo | The *code* version for Alembic and the *db* version for sqla-migrate ... | |
| 08:56:48 | bauzas | we recently packed the versions | |
| 08:56:53 | bauzas | https://github.com/openstack/nova/tree/master/nova/db/main/migrations/versions | |
| 08:56:58 | bauzas | for the cell DB | |
| 08:57:12 | bauzas | and https://github.com/openstack/nova/tree/master/nova/db/api/migrations/versions for the API DB one | |
| 08:57:33 | bauzas | HTH | |
| 08:57:41 | bauzas | (if I understood correctly your question) | |
| 09:02:34 | zigo | I'm not explaining what I need correctly I guess. My question is: | |
| 09:02:34 | zigo | - for alembic: how to find out what's the latest revision ID in /usr/lib/python3/dist-packges/nova/db ? | |
| 09:02:34 | zigo | - for sqla-migrate: how to find out the currently installed revision of the db in /var/lib/mysql ? | |
| 09:04:04 | bauzas | ah I see | |
| 09:04:49 | bauzas | well, I don't think we have some file saying in some variable exactly which version is the last, it's automatically calculed by alembic | |
| 09:05:06 | bauzas | but when you look at the files in github, you can find the last commit date | |
| 09:05:32 | bauzas | then, if you open one of the version (like the last committed one), you find its version + the parent one | |
| 09:07:06 | bauzas | https://github.com/openstack/nova/blob/master/nova/db/main/migrations/versions/1acf2c98e646_add_compute_id_to_instance.py#L25-L26 | |
| 09:07:13 | bauzas | that's the last revision | |
| 09:08:11 | frickler | zigo: does https://alembic.sqlalchemy.org/en/latest/tutorial.html#getting-information help? | |
| 09:08:14 | bauzas | but I guess the simpliest is calling the alembic command in the repo | |
| 09:08:51 | bauzas | yeah, "alembic current" will give you this | |
| 09:09:43 | zigo | Oh, I just found out for migrate ... :) | |
| 09:10:39 | zigo | root@controller-1 /usr/lib/python3/dist-packages/nova/db/sqlalchemy/migrate_repo # python3 manage.py db_version --url="mysql+pymysql://nova:<password>@<ip>/novadb?charset=utf8" | |
| 09:15:32 | zigo | root@C1-cl1-controller-1>_ /usr/lib/python3/dist-packages/nova/db/main # alembic -c alembic2.ini current | |
| 09:15:33 | zigo | INFO [alembic.runtime.migration] Context impl MySQLImpl. | |
| 09:15:33 | zigo | INFO [alembic.runtime.migration] Will assume non-transactional DDL. | |
| 09:15:33 | zigo | ERROR [alembic.util.messaging] Can't locate revision identified by 'cdeec0c85668' | |
| 09:15:33 | zigo | FAILED: Can't locate revision identified by 'cdeec0c85668' | |
| 09:16:12 | zigo | (I replaced sqlalchemy.url= in this alembic2.ini, as the original one doesn't contain the connection info as per nova.conf...) | |
| 09:16:21 | zigo | Oh, my bad. | |
| 09:16:27 | zigo | novapidb vs novadb ... | |
| 09:17:16 | zigo | Thanks a lot ! | |
| 11:41:59 | opendevreview | Pavlo Shchelokovskyy proposed openstack/nova master: Do not store InstanceInvalidState as instance_fault https://review.opendev.org/c/openstack/nova/+/860702 | |
| 12:15:13 | noonedeadpunk | hey everyone. I have weird issues with castellan when attaching encrypted volumes. So it fails on nova-compute on attach attempt with smth like this: https://paste.openstack.org/show/bLY6qE339z2dD5eZntFM/ | |
| 12:15:37 | noonedeadpunk | We've start catching that right after upgrade to 2023.1 | |
| 12:16:50 | noonedeadpunk | I'm just not sure really where to start looking into - if it's smth related to nova or keystone or castellan itself... So just checking if you might heard about smth like that before | |
| 12:19:18 | frickler | noonedeadpunk: which version of castellan are you using? there were some last-minute reverts for 2023.2 because of compatibility issues https://review.opendev.org/q/topic:revert-castellan-bobcat | |
| 12:19:42 | frickler | not sure though whether it is really the same issue you are seeing | |
| 12:22:59 | noonedeadpunk | let me see.... | |
| 12:23:21 | noonedeadpunk | As I was looking into secret users already but not sure that's it | |
| 12:24:10 | frickler | hmm, the errors in https://review.opendev.org/c/openstack/requirements/+/883141 look different, but that's for func tests only | |
| 12:27:49 | noonedeadpunk | 4.1.0 | |
| 12:28:08 | noonedeadpunk | so it shouldn't have any of these I assume | |
| 12:28:20 | noonedeadpunk | (some win from u-c usage at least | |
| 12:33:09 | noonedeadpunk | I have some feeling it is related to service tokens... But visually everything is in place... | |
| 12:33:41 | noonedeadpunk | as it fails in self.service_auth.get_token | |
| 12:33:53 | noonedeadpunk | (for keystoneauth1) | |
| 12:42:58 | noonedeadpunk | `service_user` section is in place, with all auth and `send_service_user_token = True`, service_token_roles_required = True in keystone_authtoken... | |
| 13:06:25 | zigo | FYI, as an operator, we are SUPER HAPPY about the BDM sync feature of Bobcat, thanks to whoever worked on that, it's going to make our lives a way less miserable !!! :) | |
| 13:07:04 | zigo | We had so many issues with the attachments in nova/cinder not being in sync ... | |
| 13:07:26 | zigo | So: thanks... ! | |
| 13:13:57 | sahid | o/ | |
| 13:13:59 | noonedeadpunk | Ok, my issue was related to missing barbican_service_user section... | |
| 13:14:05 | sahid | hey bauzas thanks for your comment on https://review.opendev.org/c/openstack/nova/+/896512 | |
| 13:14:22 | sahid | I will create a launchpad bp and update the change | |
| 13:14:26 | bauzas | ++ | |
| 13:14:47 | bauzas | sorry for the paperwork but I wanted to have it tracked correctly for the operators | |
| 13:15:18 | bauzas | looks to me more a feature than a bugfix | |
| 13:15:26 | bauzas | (but I could be wrong) | |
| 13:15:54 | sahid | I don't know, it's an improvment I guess | |
| 13:16:26 | sahid | that fixes a bug reported | |
| 13:16:45 | sahid | s/bug/limitation | |
| 13:17:17 | bauzas | yup, hence my point | |
| 13:17:32 | bauzas | anyway, we'll see what others say | |
| 13:18:03 | bauzas | sahid: if you create a blueprint, could you please be in a next nova meeting for discussing it ? | |
| 13:18:19 | bauzas | (Tuesdays 16UTC) | |
| 13:39:54 | dansmith | bauzas: I think you owe me a response on my alias spec about mdevs? specifically about mdevs not yet being attachable at runtime (like ports and disks)? | |
| 13:40:13 | bauzas | did I forget anything ? shit if so | |
| 13:40:33 | bauzas | -ETOOMANYGERRITEMAILS | |
| 13:41:02 | dansmith | bauzas: well you left it as "yeah then we can do mdevs too" and I said "I want to do all attachable devices, but I think that's just disks and ports" and then we had a private convo about it | |
| 13:41:24 | dansmith | and I think we decided that it was probably just ports and disks in that category, but you said you'd circle back | |
| 13:42:07 | dansmith | unrelated, gibi I wrote up this patch we discussed, can you have a look? https://review.opendev.org/c/openstack/nova/+/897076 | |
| 13:42:54 | dansmith | gibi: there's another assertion patch above that that came from a real-world situation and I think it'd probably be good to have that in a scenario where you're moving cell databases around :) | |
| 13:43:07 | bauzas | dansmith: okidoki, so I'll add a comment and review it again | |
| 13:43:37 | dansmith | bauzas: well, I just want to close the loop so it doesn't look like it's waiting for a change :) | |
| 13:45:06 | bauzas | dansmith: correct, hence my new comment needed :) | |
| 13:45:23 | bauzas | you explained me that it would only be for attach/detach calls | |
| 13:45:30 | bauzas | that mdevs don't do | |
| 13:45:47 | bauzas | that said I said I should look at other devices, if we do this | |
| 13:45:59 | bauzas | but I think I found none of them | |
| 13:46:04 | dansmith | yep, cool :) | |
| 14:12:26 | dansmith | bauzas: thanks for that | |
| 14:13:05 | gibi | dansmith: thanks I will take a look | |
| 14:13:11 | dansmith | gibi: thanks | |
| 14:24:18 | sahid | bauzas: sure I will do | |
| 17:23:58 | opendevreview | sean mooney proposed openstack/nova master: add initial healthcheck support https://review.opendev.org/c/openstack/nova/+/825015 | |
| 17:23:58 | opendevreview | sean mooney proposed openstack/nova master: add healthcheck manager to manager base https://review.opendev.org/c/openstack/nova/+/827844 | |