Search before asking
What happened
I find when there is workflowInstance running, and the database restart. The workflow instance's status will be incorrect, it will be running all the time. This is due to when we receive the worker's response, we will update the taskInstance's meta, and update the database, but when we update database failed, we will not rollback the meta in memory, so the worker send response again, the master will skip the state, due to it think the state has already been handled.
What you expected to happen
The master can recover, when database restart.
How to reproduce
- Running 1000 workflowIntsance
- Restart database
- Find some workflowInstance will always be running
Anything else
No response
Version
dev
Are you willing to submit PR?
Code of Conduct