Currently, the region failover procedure has follow issues:
The region failover doesn't support migrating a region without data loss.
The failover detector relies on the receiving heartbeats from the datanode; if the datanode doesn't send any heartbeats, we can't detect the failed region.
We can't track the in-flight failover procedures, which means we can't trigger the failover procedure again if a failover procedure fails.
What type of enhancement is this?
Refactor, Tech debt reduction
What does the enhancement do?
Currently, the region failover procedure has follow issues:
Implementation challenges
No response