[HADOOP-6762] exception while doing RPC I/O closes channel

• Type: Bug
• Status: Closed
• Priority: Critical
• Resolution: Fixed
• Affects Version/s: 0.20.2
• Fix Version/s: 2.0.3-alpha
• Target Version/s: 2.0.3-alpha
• Hadoop Flags: Reviewed
• Assignee: sam rash
• Reporter: sam rash
• Created: 12/May/10 16:24
• Updated: 30/Jun/22 22:57
• Resolved: 10/Dec/12 21:26

Description

If a single process creates two unique fileSystems to the same NN using FileSystem.newInstance(), and one of them issues a close(), the leasechecker thread is interrupted. This interrupt races with the rpc namenode.renew() and can cause a ClosedByInterruptException. This closes the underlying channel and the other filesystem, sharing the connection will get errors.

Issue Links

• causes: HADOOP-18324 Interrupting RPC Client calls can lead to thread exhaustion
• is related to: HADOOP-6768 RPC client can response more efficiently when sendParam() got IOException
• relates to: HADOOP-9107 Hadoop IPC client eats InterruptedException and sets interrupt on the thread which is not documented
• relates to: HIVE-4436 hive.exec.parallel=true doesn't work on hadoop-2
• relates to: MAPREDUCE-2392 TaskTracker shutdown in the tests sometimes take 60s