Fix ansible collections import error #1987
Conversation
Avoid "duplicate key value violates unique" error when importing ansible collections. fixes pulp#1986
| instance = super().get_instance(instance_loader, row) | ||
| if not instance and row["is_highest"] == "1": | ||
| CollectionVersion.objects.filter( | ||
| collection_id=row["collection"], is_highest=True | ||
| ).update(is_highest=False) | ||
|
|
||
| return instance |
There was a problem hiding this comment.
Oh damn it! is_highest really needs to go away.
Thank you for finding this.
I think however, we cannot be sure that when importing a highest collection version, that there is no higher one in the system already. Am I right, that you don't actually care about the is_highest flag? And this seems to fix the issue with the database constraint?
I guess we could also just lift the constraint, if we realize we want to accept inconsistent data here.
|
Speaking with the ansible folks, they don't seem to care about that field anymore either. |
|
Hey, can you have a try if this magically solves the issue for you too? |
|
Yeah, I'm thinking backporting a fix that removes the field from the api's is probably a bad thing to do. |
Not really. As far as I see, it will run only one additional sql update query per collection.
I think it is ok if we are sure we don't need it? I am not exactly sure what is depending on this flag. I didn't check if Satellite needs it or not.
Satellite 6.15 (pulp ansible 0.20.8) and Satellite 6.16 (pulp ansible 0.21.8 i think) ? |
Thank you. (So 0.20 and 0.21 branches...) |
|
This is my proposal for a fix. WDYT? |
Avoid "duplicate key value violates unique" error
when importing ansible collections.
fixes #1986