[KAFKA-9207] Replica Out of Sync as creating ReplicaFetcher thread failed with connection to leader

• Type: Bug
• Status: Open
• Priority: Major
• Resolution: Unresolved
• Affects Version/s: 2.3.0
• Component/s: replication
• Assignee: Unassigned
• Reporter: Xue Liu
• Created: 18/Nov/19 19:57
• Updated: 18/Nov/19 23:38

Description

We sometimes see a replica for a partition is out of sync. When the issue happens, it seems that we just lost that replica (would never catch-up), unless we restart that broker.

It appears that ReplicaFetcher thread for that partition is dead and broker will not restart that thread. We didn't see any exception in server or controller logs.

The screen capture is taken from the broker that has that replica. Leader is 2017.