In case kafka supervisor spec metadata insertion fails then in the response database exception is returned. However since the supervisor is started before metadata insertion the indexing tasks continues to run even though the insertion failed. I think in case of metadata failures the supervisor should just stop.
This happened when I posted a supervisor spec to overlord but the supervisor table schema in the database had version field however I checked recently the schema has been changed to have created_date. Therefore, I had to delete the supervisor table and restart the overlord to have the new table schema.
In case kafka supervisor spec metadata insertion fails then in the response database exception is returned. However since the supervisor is started before metadata insertion the indexing tasks continues to run even though the insertion failed. I think in case of metadata failures the supervisor should just stop.
This happened when I posted a supervisor spec to overlord but the supervisor table schema in the database had
versionfield however I checked recently the schema has been changed to havecreated_date. Therefore, I had to delete the supervisor table and restart the overlord to have the new table schema.