Package org.apache.hadoop.hbase.regionserver.regionreplication

Class RegionReplicationFlushRequester

java.lang.Object

org.apache.hadoop.hbase.regionserver.regionreplication.RegionReplicationFlushRequester

@Private
class RegionReplicationFlushRequester
extends Object

A helper class for requesting flush on a given region.

In general, we do not want to trigger flush too frequently for a region, so here we will add something like a rate control, i.e, the interval of the two flush request should not be too small.

Field Summary

Fields

Modifier and Type	Field	Description
private final Runnable	flushRequester
private long	lastFlushedSequenceId
private long	lastRequestNanos
static final String	MIN_INTERVAL_SECS	The minimum interval between two flush requests
static final int	MIN_INTERVAL_SECS_DEFAULT
private final long	minIntervalSecs
private org.apache.hbase.thirdparty.io.netty.util.Timeout	pendingFlushRequest
private long	pendingFlushRequestSequenceId
private static org.apache.hbase.thirdparty.io.netty.util.HashedWheelTimer	TIMER	The timer for executing delayed flush request task.

Constructor Summary

Constructors

Constructor	Description
RegionReplicationFlushRequester(org.apache.hadoop.conf.Configuration conf, Runnable flushRequester)

Method Summary

Modifier and Type	Method	Description
private void	flush(org.apache.hbase.thirdparty.io.netty.util.Timeout timeout)
private static org.apache.hbase.thirdparty.io.netty.util.HashedWheelTimer	getTimer()
(package private) void	recordFlush(long sequenceId)	Record that we have already finished a flush with the given sequenceId.
private void	request()
(package private) void	requestFlush(long sequenceId)	Request a flush for the given region.

Methods inherited from class java.lang.Object

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

Field Details

TIMER

private static volatile org.apache.hbase.thirdparty.io.netty.util.HashedWheelTimer TIMER

The timer for executing delayed flush request task.

It will be shared across all the instances RegionReplicationFlushRequester. Created on demand to save one extra thread as not every user uses region replication.

MIN_INTERVAL_SECS

public static final String MIN_INTERVAL_SECS

The minimum interval between two flush requests

See Also:

• Constant Field Values

MIN_INTERVAL_SECS_DEFAULT

public static final int MIN_INTERVAL_SECS_DEFAULT

See Also:

• Constant Field Values

flushRequester

private final Runnable flushRequester

minIntervalSecs

private final long minIntervalSecs

lastRequestNanos

private long lastRequestNanos

pendingFlushRequestSequenceId

private long pendingFlushRequestSequenceId

lastFlushedSequenceId

private long lastFlushedSequenceId

pendingFlushRequest

private org.apache.hbase.thirdparty.io.netty.util.Timeout pendingFlushRequest

Constructor Details

RegionReplicationFlushRequester

RegionReplicationFlushRequester(org.apache.hadoop.conf.Configuration conf,
 Runnable flushRequester)

Method Details

getTimer

private static org.apache.hbase.thirdparty.io.netty.util.HashedWheelTimer getTimer()

request

private void request()

flush

private void flush(org.apache.hbase.thirdparty.io.netty.util.Timeout timeout)

requestFlush

void requestFlush(long sequenceId)

Request a flush for the given region.

The sequence id of the edit which we fail to replicate. A flush must happen after this sequence id to recover the failure.

recordFlush

void recordFlush(long sequenceId)

Record that we have already finished a flush with the given sequenceId.

We can cancel the pending flush request if the failed sequence id is less than the given sequenceId.