[HBASE-11145] UNEXPECTED!!! when HLog sync: Queue full

• Type: Bug
• Status: Closed
• Priority: Critical
• Resolution: Fixed
• Fix Version/s: 0.99.1
• Component/s: wal
• Hadoop Flags: Reviewed
• Assignee: Michael Stack
• Reporter: Anoop Sam John
• Created: 09/May/14 05:32
• Updated: 06/Apr/18 17:55
• Resolved: 19/Sep/14 21:32

Description

Got the below Exceptions Log in case of a write heavy test

2014-05-07 11:29:56,417 ERROR [main.append-pool1-t1] wal.FSHLog$RingBufferEventHandler(1882): UNEXPECTED!!!
java.lang.IllegalStateException: Queue full
 at java.util.AbstractQueue.add(Unknown Source)
 at org.apache.hadoop.hbase.regionserver.wal.FSHLog$SyncRunner.offer(FSHLog.java:1227)
 at org.apache.hadoop.hbase.regionserver.wal.FSHLog$RingBufferEventHandler.onEvent(FSHLog.java:1878)
 at org.apache.hadoop.hbase.regionserver.wal.FSHLog$RingBufferEventHandler.onEvent(FSHLog.java:1)
 at com.lmax.disruptor.BatchEventProcessor.run(BatchEventProcessor.java:133)
 at java.util.concurrent.ThreadPoolExecutor$Worker.runTask(Unknown Source)
 at java.util.concurrent.ThreadPoolExecutor$Worker.run(Unknown Source)
 at java.lang.Thread.run(Unknown Source)
2014-05-07 11:29:56,418 ERROR [main.append-pool1-t1] wal.FSHLog$RingBufferEventHandler(1882): UNEXPECTED!!!
java.lang.ArrayIndexOutOfBoundsException: 5
 at org.apache.hadoop.hbase.regionserver.wal.FSHLog$RingBufferEventHandler.onEvent(FSHLog.java:1838)
 at org.apache.hadoop.hbase.regionserver.wal.FSHLog$RingBufferEventHandler.onEvent(FSHLog.java:1)
 at com.lmax.disruptor.BatchEventProcessor.run(BatchEventProcessor.java:133)
 at java.util.concurrent.ThreadPoolExecutor$Worker.runTask(Unknown Source)
 at java.util.concurrent.ThreadPoolExecutor$Worker.run(Unknown Source)
 at java.lang.Thread.run(Unknown Source)
2014-05-07 11:29:56,419 ERROR [main.append-pool1-t1] wal.FSHLog$RingBufferEventHandler(1882): UNEXPECTED!!!
java.lang.ArrayIndexOutOfBoundsException: 6
 at org.apache.hadoop.hbase.regionserver.wal.FSHLog$RingBufferEventHandler.onEvent(FSHLog.java:1838)
 at org.apache.hadoop.hbase.regionserver.wal.FSHLog$RingBufferEventHandler.onEvent(FSHLog.java:1)
 at com.lmax.disruptor.BatchEventProcessor.run(BatchEventProcessor.java:133)
 at java.util.concurrent.ThreadPoolExecutor$Worker.runTask(Unknown Source)
 at java.util.concurrent.ThreadPoolExecutor$Worker.run(Unknown Source)
 at java.lang.Thread.run(Unknown Source)
2014-05-07 11:29:56,419 ERROR [main.append-pool1-t1] wal.FSHLog$RingBufferEventHandler(1882): UNEXPECTED!!!
java.lang.ArrayIndexOutOfBoundsException: 7
 at org.apache.hadoop.hbase.regionserver.wal.FSHLog$RingBufferEventHandler.onEvent(FSHLog.java:1838)
 at org.apache.hadoop.hbase.regionserver.wal.FSHLog$RingBufferEventHandler.onEvent(FSHLog.java:1)
 at com.lmax.disruptor.BatchEventProcessor.run(BatchEventProcessor.java:133)
 at java.util.concurrent.ThreadPoolExecutor$Worker.runTask(Unknown Source)
 at java.util.concurrent.ThreadPoolExecutor$Worker.run(Unknown Source)
 at java.lang.Thread.run(Unknown Source)

 

In FSHLog$SyncRunner.offer we do BlockingQueue.add() which throws Exception as it is full. The problem is the below shown catch() we do not do any cleanup.

this.syncRunners[index].offer(sequence, this.syncFutures, this.syncFuturesCount);
        attainSafePoint(sequence);
        this.syncFuturesCount = 0;
      } catch (Throwable t) {
        LOG.error("UNEXPECTED!!!", t);
      }

syncFuturesCount is not getting reset to 0 and so the subsequent onEvent() handling throws ArrayIndexOutOfBoundsException.

I think we should do the below
1. Handle the Exception and call cleanupOutstandingSyncsOnException() as in other cases of Exception handling
2. Instead of BlockingQueue.add() use offer()