Details
-
Type:
Bug
-
Status: Done
-
Resolution: Done
-
Fix Version/s: None
-
Component/s: daf_butler
-
Labels:
-
Story Points:1
-
Epic Link:
-
Sprint:BG3_F18_10
-
Team:Data Access and Database
Description
My first attempt to implement constraint check in associate() method (DM-15686) apparently broke everything (DM-16159). Looks like code in obs_subaru depends on uniqueness check to be done in addDataset() method. My commit moved that check to associate() which is apparently too late, tha fix for that it to leave the test in addDataset(). IT would be interesting to also understand what is going on in obs_subaru because in the future we'll probably switch to table-level constraints and that can break things again.
I agree 100% with all of that (both diagnosis and workaround). I suspect we will need to start using explicit ON CONFLICT instructions or something, or perhaps scale back what we think we can promise in terms of guarantees in ingest.