org.apache.hadoop.hbase.regionserver

Class TestWALLockup

java.lang.Object

org.apache.hadoop.hbase.regionserver.TestWALLockup

public class TestWALLockup
extends Object

Testing for lock up of WAL subsystem. Copied from TestHRegion.

Field Summary

Fields

Modifier and Type	Field and Description
TestName	name
protected TableName	tableName

Constructor Summary

Constructors

Constructor and Description
TestWALLockup()

Method Summary

Modifier and Type	Method and Description
HRegion	initHRegion(TableName tableName, byte[] startKey, byte[] stopKey, WAL wal)
void	setup()
void	tearDown()
void	testLockup16960() Reproduce locking up that happens when there's no further syncs after append fails, and causing an isolated sync then infinite wait.
void	testLockupWhenSyncInMiddleOfZigZagSetup() Reproduce locking up that happens when we get an inopportune sync during setup for zigzaglatch wait.

Methods inherited from class java.lang.Object

clone, equals, finalize, getClass, hashCode, notify, notifyAll, toString, wait, wait, wait

Field Detail

name

public TestName name

tableName

protected TableName tableName

Constructor Detail

TestWALLockup

public TestWALLockup()

Method Detail

setup

public void setup()
           throws IOException

Throws:

IOException

tearDown

public void tearDown()
              throws Exception

Throws:

Exception

testLockupWhenSyncInMiddleOfZigZagSetup

public void testLockupWhenSyncInMiddleOfZigZagSetup()
                                             throws IOException

Reproduce locking up that happens when we get an inopportune sync during setup for zigzaglatch wait. See HBASE-14317. If below is broken, we will see this test timeout because it is locked up.

First I need to set up some mocks for Server and RegionServerServices. I also need to set up a dodgy WAL that will throw an exception when we go to append to it.

Throws:

IOException

testLockup16960

public void testLockup16960()
                     throws IOException

Reproduce locking up that happens when there's no further syncs after append fails, and causing an isolated sync then infinite wait. See HBASE-16960. If below is broken, we will see this test timeout because it is locked up.

Steps for reproduce:
1. Trigger server abort through dodgyWAL1
2. Add a DummyWALActionsListener to dodgyWAL2 to cause ringbuffer event handler thread sleep for a while thus keeping endOfBatch false
3. Publish a sync then an append which will throw exception, check whether the sync could return

Throws:

IOException

initHRegion

public HRegion initHRegion(TableName tableName,
                           byte[] startKey,
                           byte[] stopKey,
                           WAL wal)
                    throws IOException

Returns:

A region on which you must call HBaseTestingUtility.closeRegionAndWAL(HRegion) when done.

Throws:

IOException