getredash
Data source groups aren't checked for uniqueness
We ran into an issue with our groups not showing any data sources. I traced it to our GET /api/groups/{id}/data_sources call returning a 500. Sentry records an error (really helpful, btw):
MultipleResultsFound
Multiple rows were found for one()
on this line:
https://github.com/getredash/redash/blob/5b54a777d91e18398f68fcae4bdc669f438faec0/redash/models.py#L491
And sure enough, we had multiple rows in data_source_groups for a single (data_source_id, group_id) pair.
I traced it back to the API for creating data source group associations:
https://github.com/getredash/redash/blob/5b54a777d91e18398f68fcae4bdc669f438faec0/redash/handlers/groups.py#L130-L131
Which doesn't check if a matching data source group row exists, and can be triggered by repeated requests if you click to add a data source multiple times or in different browsers.
What do you think about adding a unique index to (data_source_id, group_id), which would prevent the duplicate data in the DB? We can also add a guard to check for an existing row and return that instead of creating a new one. Checking only in Python, though, still leaves us open to a race condition on the write.
Extra interesting because DELETE /api/groups/{groupId}/data_sources/{dataSourceId} will delete all duplicates, but GET returns 500 as you reported 🤷
