org.apache.hadoop.hbase.regionserver

Class TestRegionReplicaFailover

java.lang.Object

org.apache.hadoop.hbase.regionserver.TestRegionReplicaFailover

public class TestRegionReplicaFailover
extends Object

Tests failover of secondary region replicas.

Field Summary

Fields

Modifier and Type	Field and Description
static HBaseClassTestRule	CLASS_RULE
protected byte[]	fam
protected byte[][]	families
private org.apache.hadoop.hbase.HTableDescriptor	htd
private static HBaseTestingUtility	HTU
private static org.slf4j.Logger	LOG
org.junit.rules.TestName	name
private static int	NB_SERVERS
protected byte[]	qual1
protected byte[]	row
protected byte[]	row2
protected byte[]	value1

Constructor Summary

Constructors

Constructor and Description
TestRegionReplicaFailover()

Method Summary

Modifier and Type	Method and Description
void	after()
void	before()
void	testLotsOfRegionReplicas() Tests the case where we are creating a table with a lot of regions and replicas.
void	testPrimaryRegionKill() Tests the case where killing a primary region with unflushed data recovers
void	testSecondaryRegionKill() Tests the case where killing a secondary region with unflushed data recovers, and the replica becomes available to read again shortly.
void	testSecondaryRegionKillWhilePrimaryIsAcceptingWrites() Tests the case where there are 3 region replicas and the primary is continuously accepting new writes while one of the secondaries is killed.
void	testSecondaryRegionWithEmptyRegion() Tests the case where a newly created table with region replicas and no data, the secondary region replicas are available to read immediately.
void	testSecondaryRegionWithNonEmptyRegion() Tests the case where if there is some data in the primary region, reopening the region replicas (enable/disable table, etc) makes the region replicas readable.
private void	verifyNumericRowsWithTimeout(org.apache.hadoop.hbase.client.Table table, byte[] f, int startRow, int endRow, int replicaId, long timeout) wal replication is async, we have to wait until the replication catches up, or we timeout

Methods inherited from class java.lang.Object

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

Field Detail

CLASS_RULE

public static final HBaseClassTestRule CLASS_RULE

LOG

private static final org.slf4j.Logger LOG

HTU

private static final HBaseTestingUtility HTU

NB_SERVERS

private static final int NB_SERVERS

See Also:

Constant Field Values

families

protected final byte[][] families

fam

protected final byte[] fam

qual1

protected final byte[] qual1

value1

protected final byte[] value1

row

protected final byte[] row

row2

protected final byte[] row2

name

public org.junit.rules.TestName name

htd

private org.apache.hadoop.hbase.HTableDescriptor htd

Constructor Detail

TestRegionReplicaFailover

public TestRegionReplicaFailover()

Method Detail

before

public void before()
            throws Exception

Throws:

Exception

after

public void after()
           throws Exception

Throws:

Exception

testSecondaryRegionWithEmptyRegion

public void testSecondaryRegionWithEmptyRegion()
                                        throws IOException

Tests the case where a newly created table with region replicas and no data, the secondary region replicas are available to read immediately.

Throws:

IOException

testSecondaryRegionWithNonEmptyRegion

public void testSecondaryRegionWithNonEmptyRegion()
                                           throws IOException

Tests the case where if there is some data in the primary region, reopening the region replicas (enable/disable table, etc) makes the region replicas readable.

Throws:

IOException

testPrimaryRegionKill

public void testPrimaryRegionKill()
                           throws Exception

Tests the case where killing a primary region with unflushed data recovers

Throws:

Exception

verifyNumericRowsWithTimeout

private void verifyNumericRowsWithTimeout(org.apache.hadoop.hbase.client.Table table,
                                          byte[] f,
                                          int startRow,
                                          int endRow,
                                          int replicaId,
                                          long timeout)
                                   throws Exception

wal replication is async, we have to wait until the replication catches up, or we timeout

Throws:

Exception

testSecondaryRegionKill

public void testSecondaryRegionKill()
                             throws Exception

Tests the case where killing a secondary region with unflushed data recovers, and the replica becomes available to read again shortly.

Throws:

Exception

testSecondaryRegionKillWhilePrimaryIsAcceptingWrites

public void testSecondaryRegionKillWhilePrimaryIsAcceptingWrites()
                                                          throws Exception

Tests the case where there are 3 region replicas and the primary is continuously accepting new writes while one of the secondaries is killed. Verification is done for both of the secondary replicas.

Throws:

Exception

testLotsOfRegionReplicas

public void testLotsOfRegionReplicas()
                              throws IOException

Tests the case where we are creating a table with a lot of regions and replicas. Opening region replicas should not block handlers on RS indefinitely.

Throws:

IOException