Details
-
Story
-
Status: Resolved
-
Major
-
Resolution: Fixed
-
None
-
None
-
None
Description
Here is a sequence of steps that can lead of a crasher:
1. Deploy the scheduler and take a snapshot with the dbScript field populated.
2. Upgrade the scheduler with a new version which contains schema changes (such as joshua.cohen's recent change to store Docker and AppC image ids.
3. After upgrading, the scheduler will restore the schema of the in memory database from the dbScript field and not the built in schema. This means the table additions from the previous step will not exist.
4. Observe crashes as queries to the in memory database.