[RATIS-1401] Failed to add a new group due to RaftServerProxy$ImplMap LOCK unavailable

• Type: Bug
• Status: Resolved
• Priority: Blocker
• Resolution: Fixed
• Fix Version/s: 2.2.0
• Component/s: server
• Assignee: Tsz-wo Sze
• Reporter: Sammi Chen
• Created: 07/Sep/21 12:21
• Updated: 10/Sep/21 15:38
• Resolved: 10/Sep/21 15:38

Description

The DN is found no response to the /jmx request. And with further investigation, this DN failed to join the new pipeline at the same time.

Attached is the jstack, it seems that DN is waiting for the RaftServerProxy$ImplMap LOCK to add new group, while the LOCK is hold by another thread to remove the old group, and this thread wants to have the readLock of it's SegmentLock, the readLock is hold by another writeChunk thread, which is waiting stateMachineDataCache quota.

See the attached jstack.

Issue Links

• links to: GitHub Pull Request #498