[HBASE-5025] Two threads named pool-1-thread-1 and pool-2-thread-1 remains after the shutdown of a cluster

• Type: Bug
• Status: Closed
• Priority: Minor
• Resolution: Not A Problem
• Affects Version/s: 0.94.0
• Component/s: IPC/RPC, test
• Assignee: Unassigned
• Reporter: Nicolas Liochon
• Created: 14/Dec/11 13:34
• Updated: 12/Jun/22 19:53
• Resolved: 11/Apr/15 01:00

Description

There are two issues with these threads:

• there should have a better name. pool-x-thread-y is the default name given by java.util.concurrent.Executors
• these threads should not survive to a minicluster shutdown

They are created by org.apache.hadoop.ipc.Server$Listener (with version 0.20.205.0; the code was different before), so the first issue is a hadoop common one. It's unclear for the second one, it could be hadoop-common as well.

Constructor for org.apache.hadoop.ipc.Server$Listener:

    public Listener() throws IOException {
      //...
      readPool = Executors.newFixedThreadPool(readThreads);  // Lack a ThreadFactory to set the names
      //...
    }

Server#stop shutdowns the thread pool.

Issue Links

• relates to: HBASE-4602 Make the suite run in at least half the time