Details
-
Bug
-
Status: Resolved
-
Minor
-
Resolution: Fixed
-
Auto failover (HDFS-3042)
-
None
-
Reviewed
Description
This test starts a cluster, then starts the ZKFCs, then tries to issue a manual failover. But, it doesn't wait until the ZKFCs are fully done starting up. So, I see a test run occasionally fail because the failover says that the NN is not ready yet. The test just needs to wait until the ZKFCs are in "healthy" state before proceeding.