Appears there might be a race condition where RecordValidations are removed from Mongo in tasks.py, but Spark immediately queries and finds they still exist, which throws off the re-write and does not update validity of Records.
If no other validations run, Job will show that no Records are invalid, but that's not the case.
Appears there might be a race condition where
RecordValidations
are removed from Mongo intasks.py
, but Spark immediately queries and finds they still exist, which throws off the re-write and does not update validity of Records.If no other validations run, Job will show that no Records are invalid, but that's not the case.