awx-migration-24.3.1 failing about awx HOT 7 OPEN

Note the last upgrade to 24.0.0 appeared to run the migration pod with no issues.

migration-24.0.0:
Operations to perform: Apply all migrations: auth, conf, contenttypes, dab_resource_registry, main, oauth2_provider, sessions, sites, social_django, sso Running migrations: Applying dab_resource_registry.0001_initial... OK Applying dab_resource_registry.0002_remove_resource_id... OK Applying dab_resource_registry.0003_alter_resource_object_id... OK Applying main.0190_alter_inventorysource_source_and_more... OK

Also tested a brand new deployment via AWX operator and latest version deploys with no issues.

from awx.

is this problem still reproducible in your environment? do you have a backup of your database pre-upgrade?

from awx.

I'd expect this message to happen if somehow the unique constraint for the team model's id field got messed up. This is happening as a ForeignKey is added to the team model, but this should not use the name field, which is not unique. Maybe somehow some change made it think the name field is the primary_key.

from awx.

is this problem still reproducible in your environment? do you have a backup of your database pre-upgrade?

Yes it is. I was able to restore from backup and get my instance upgraded to 24.2.0 however it appears even with the version i am on trying to upgrade to 24.3.1 has the same issue.

from awx.

I'd expect this message to happen if somehow the unique constraint for the team model's id field got messed up. This is happening as a ForeignKey is added to the team model, but this should not use the name field, which is not unique. Maybe somehow some change made it think the name field is the primary_key.

Interesting, any ideas on how to resolve? I'm looking at the postgres db and i see the table in question but nothing looks out of the ordinary. Could i manually set the primary_key back to id? Interesting thing is i don't see a unique constraint listed at the bottom of the table output.

Another fresh install of AWX main_team table for reference:

from awx.

FYI - I "resolved" the issue by following these steps. Probably not the cleanest way but i needed to get my instance up and running with the latest release.

1. psql dump fresh install of AWX postgres DB table main_team pg_dump --table main_team awx > main_team_bkp
2. oc rsh into postgres pod on broken instance oc project ansible-awx oc rsh awx-prod-postgres-15-0
3. take psql dump backup of existing DB (the one having issues) pg_dump awx > awx_bkp
4. import psql dump of fresh main_team table psql awx < main_team_bkp
5. perform awx-manage migrate command on web pod
6. reload web/task pods and everything seems happy now.

Any thoughts on if this is a terrible idea or if this should be ok?

from awx.

I'm almost completely convinced that something messed up the indices of your team table. I think you have a great strategy for the immediate issue. My only concern would be whether the constraints of other tables also got corrupted. Looking into the obvious tools for this, it seems we already have django-extensions installed so awx-manage sqldiff main dab_rbac should work, but it gives a lot of junk output, here is what I get in a fresh DB:

https://gist.github.com/AlanCoding/dcd9e67e02423e5524450b8150a4d6d8

So you could run that and cross-reference against mine. If you have other tables which dropped constraints, I wonder if it might still be obvious enough you can compare and see it.

from awx.