[HBASE-9796] npe in RegionServerCallable

• Type: Bug
• Status: Closed
• Priority: Major
• Resolution: Fixed
• Affects Version/s: 0.96.0
• Fix Version/s: 0.98.0, 0.96.1
• Hadoop Flags: Reviewed
• Assignee: Michael Stack
• Reporter: Michael Stack
• Created: 17/Oct/13 22:03
• Updated: 16/Dec/13 18:46
• Resolved: 18/Oct/13 00:09

Description

Saw this running hbase-it suite on test cluster. Its tricky. Needs a little study. Connection or location became null in this code when we go to clear the caches:

    if (t instanceof SocketTimeoutException ||
        t instanceof ConnectException ||
        t instanceof RetriesExhaustedException ||
        (location != null && getConnection().isDeadServer(location.getServerName()))) {
      // if thrown these exceptions, we clear all the cache entries that
      // map to that slow/dead server; otherwise, let cache miss and ask
      // hbase:meta again to find the new location
      getConnection().clearCaches(location.getServerName());

Here is exception seen:

2013-10-17 09:05:44,569 INFO [Thread-9] actions.Action: Killed region server:a1811.halxg.cloudera.com,60020,1382025707403. Reported num of rs:4
2013-10-17 09:05:44,569 INFO [Thread-9] actions.Action: Sleeping for:2961
Exception in thread "HBaseUpdaterThread_3" java.lang.NullPointerException
at org.apache.hadoop.hbase.client.RegionServerCallable.throwable(RegionServerCallable.java:120)
at org.apache.hadoop.hbase.client.RpcRetryingCaller.callWithRetries(RpcRetryingCaller.java:124)
at org.apache.hadoop.hbase.client.HTable.get(HTable.java:755)
at org.apache.hadoop.hbase.util.MultiThreadedUpdater$HBaseUpdaterThread.run(MultiThreadedUpdater.java:159)