org.apache.hadoop.hbase.master.zksyncer

Class ClientZKSyncer

java.lang.Object

org.apache.hadoop.hbase.zookeeper.ZKListener

org.apache.hadoop.hbase.master.zksyncer.ClientZKSyncer

Direct Known Subclasses:

MasterAddressSyncer, MetaLocationSyncer

@InterfaceAudience.Private
public abstract class ClientZKSyncer
extends ZKListener

Tracks the target znode(s) on server ZK cluster and synchronize them to client ZK cluster if changed

The target znode(s) is given through getPathsToWatch() method

Nested Class Summary

Nested Classes

Modifier and Type	Class and Description
private class	ClientZKSyncer.ClientZkUpdater Thread to synchronize znode data to client ZK cluster
private static class	ClientZKSyncer.ZKData Used to store the newest data which we want to sync to client zk.

Field Summary

Fields

Modifier and Type	Field and Description
private ZKWatcher	clientZkWatcher
private static org.slf4j.Logger	LOG
private ConcurrentMap<String,ClientZKSyncer.ZKData>	queues
private Server	server

Fields inherited from class org.apache.hadoop.hbase.zookeeper.ZKListener

watcher

Constructor Summary

Constructors

Constructor and Description
ClientZKSyncer(ZKWatcher watcher, ZKWatcher clientZkWatcher, Server server)

Method Summary

Modifier and Type	Method and Description
private void	deleteDataForClientZkUntilSuccess(String node)
private void	getDataAndWatch(String path)
protected abstract Set<String>	getPathsToWatch() Returns the zk path(s) to watch
void	nodeCreated(String path) Called when a new node has been created.
void	nodeDataChanged(String path) Called when an existing node has changed data.
void	nodeDeleted(String path) Called when a node has been deleted
private void	reconnectAfterExpiration()
protected void	refreshWatchingList()
private void	removeQueue(String path)
private void	setDataForClientZkUntilSuccess(String node, byte[] data) Set data for client ZK and retry until succeed.
void	start() Starts the syncer
private void	startNewSyncThread(String path)
private void	upsertQueue(String node, byte[] data) Update the value of the single element in queue if any, or else insert.
protected abstract boolean	validate(String path) Validate whether a znode path is watched by us
private void	watchAndCheckExists(String node)

Methods inherited from class org.apache.hadoop.hbase.zookeeper.ZKListener

getWatcher, nodeChildrenChanged

Methods inherited from class java.lang.Object

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

Field Detail

LOG

private static final org.slf4j.Logger LOG

server

private final Server server

clientZkWatcher

private final ZKWatcher clientZkWatcher

queues

private final ConcurrentMap<String,ClientZKSyncer.ZKData> queues

Constructor Detail

ClientZKSyncer

public ClientZKSyncer(ZKWatcher watcher,
                      ZKWatcher clientZkWatcher,
                      Server server)

Method Detail

startNewSyncThread

private void startNewSyncThread(String path)

start

public void start()
           throws org.apache.zookeeper.KeeperException

Starts the syncer

Throws:

org.apache.zookeeper.KeeperException - if error occurs when trying to create base nodes on client ZK

watchAndCheckExists

private void watchAndCheckExists(String node)

upsertQueue

private void upsertQueue(String node,
                         byte[] data)

Update the value of the single element in queue if any, or else insert.

We only need to synchronize the latest znode value to client ZK rather than synchronize each time

Parameters:

data - the data to write to queue

setDataForClientZkUntilSuccess

private void setDataForClientZkUntilSuccess(String node,
                                            byte[] data)
                                     throws InterruptedException

Set data for client ZK and retry until succeed. Be very careful to prevent dead loop when modifying this method

Parameters:

node - the znode to set on client ZK

data - the data to set to client ZK

Throws:

InterruptedException - if the thread is interrupted during process

deleteDataForClientZkUntilSuccess

private void deleteDataForClientZkUntilSuccess(String node)
                                        throws InterruptedException

Throws:

InterruptedException

reconnectAfterExpiration

private final void reconnectAfterExpiration()
                                     throws InterruptedException

Throws:

InterruptedException

getDataAndWatch

private void getDataAndWatch(String path)

removeQueue

private void removeQueue(String path)

nodeCreated

public void nodeCreated(String path)

Description copied from class: ZKListener

Called when a new node has been created.

Overrides:

nodeCreated in class ZKListener

Parameters:

path - full path of the new node

nodeDataChanged

public void nodeDataChanged(String path)

Description copied from class: ZKListener

Called when an existing node has changed data.

Overrides:

nodeDataChanged in class ZKListener

Parameters:

path - full path of the updated node

nodeDeleted

public void nodeDeleted(String path)

Description copied from class: ZKListener

Called when a node has been deleted

Overrides:

nodeDeleted in class ZKListener

Parameters:

path - full path of the deleted node

validate

protected abstract boolean validate(String path)

Validate whether a znode path is watched by us

Parameters:

path - the path to validate

Returns:

true if the znode is watched by us

getPathsToWatch

protected abstract Set<String> getPathsToWatch()

Returns the zk path(s) to watch

refreshWatchingList

protected final void refreshWatchingList()