org.apache.hadoop.hbase.backup.example

Class TableHFileArchiveTracker

java.lang.Object

org.apache.hadoop.hbase.zookeeper.ZKListener

org.apache.hadoop.hbase.backup.example.TableHFileArchiveTracker

@InterfaceAudience.Private
public final class TableHFileArchiveTracker
extends ZKListener

Track HFile archiving state changes in ZooKeeper. Keeps track of the tables whose HFiles should be kept in the archive.

start() needs to be called to start monitoring for tables to archive.

Field Summary

Fields

Modifier and Type	Field and Description
private String	archiveHFileZNode
static String	HFILE_ARCHIVE_ZNODE_PARENT
private static org.slf4j.Logger	LOG
private HFileArchiveTableMonitor	monitor
private boolean	stopped

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

watcher

Constructor Summary

Constructors

Modifier	Constructor and Description
private	TableHFileArchiveTracker(ZKWatcher watcher, HFileArchiveTableMonitor monitor)

Method Summary

Modifier and Type	Method and Description
private void	addAndReWatchTable(String tableZnode) Add this table to the tracker and then read a watch on that node.
private void	checkEnabledAndUpdate() Sets the watch on the top-level archive znode, and then updates the monitor with the current tables that should be archived (and ensures that those nodes are watched as well).
private void	clearTables() Remove the currently archived tables.
static TableHFileArchiveTracker	create(org.apache.hadoop.conf.Configuration conf) Create an archive tracker for the passed in server
private static TableHFileArchiveTracker	create(ZKWatcher zkw, HFileArchiveTableMonitor monitor) Create an archive tracker with the special passed in table monitor.
HFileArchiveTableMonitor	getMonitor()
ZKWatcher	getZooKeeperWatcher()
boolean	keepHFiles(String tableName) Determine if the given table should or should not allow its hfiles to be deleted
void	nodeChildrenChanged(String path) Called when an existing node has a child node added or removed.
void	nodeCreated(String path) Called when a new node has been created.
void	nodeDeleted(String path) Called when a node has been deleted
private void	safeStopTrackingTable(String tableZnode) Stop tracking a table.
void	start() Start monitoring for archive updates
void	stop() Stop this tracker and the passed zookeeper
private void	updateWatchedTables() Read the list of children under the archive znode as table names and then sets those tables to the list of tables that we should archive

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

getWatcher, nodeDataChanged

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

HFILE_ARCHIVE_ZNODE_PARENT

public static final String HFILE_ARCHIVE_ZNODE_PARENT

See Also:

Constant Field Values

monitor

private HFileArchiveTableMonitor monitor

archiveHFileZNode

private String archiveHFileZNode

stopped

private boolean stopped

Constructor Detail

TableHFileArchiveTracker

private TableHFileArchiveTracker(ZKWatcher watcher,
                                 HFileArchiveTableMonitor monitor)

Method Detail

start

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

Start monitoring for archive updates

Throws:

org.apache.zookeeper.KeeperException - on failure to find/create nodes

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

nodeChildrenChanged

public void nodeChildrenChanged(String path)

Description copied from class: ZKListener

Called when an existing node has a child node added or removed.

Overrides:

nodeChildrenChanged in class ZKListener

Parameters:

path - full path of the node whose children have changed

addAndReWatchTable

private void addAndReWatchTable(String tableZnode)
                         throws org.apache.zookeeper.KeeperException

Add this table to the tracker and then read a watch on that node.

Handles situation where table is deleted in the time between the update and resetting the watch by deleting the table via safeStopTrackingTable(String)

Parameters:

tableZnode - full zookeeper path to the table to be added

Throws:

org.apache.zookeeper.KeeperException - if an unexpected zk exception occurs

safeStopTrackingTable

private void safeStopTrackingTable(String tableZnode)
                            throws org.apache.zookeeper.KeeperException

Stop tracking a table. Ensures that the table doesn't exist, but if it does, it attempts to add the table back via addAndReWatchTable(String) - its a 'safe' removal.

Parameters:

tableZnode - full zookeeper path to the table to be added

Throws:

org.apache.zookeeper.KeeperException - if an unexpected zk exception occurs

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

checkEnabledAndUpdate

private void checkEnabledAndUpdate()

Sets the watch on the top-level archive znode, and then updates the monitor with the current tables that should be archived (and ensures that those nodes are watched as well).

updateWatchedTables

private void updateWatchedTables()
                          throws org.apache.zookeeper.KeeperException

Read the list of children under the archive znode as table names and then sets those tables to the list of tables that we should archive

Throws:

org.apache.zookeeper.KeeperException - if there is an unexpected zk exception

clearTables

private void clearTables()

Remove the currently archived tables.

Does some intelligent checking to make sure we don't prematurely create an archive tracker.

keepHFiles

public boolean keepHFiles(String tableName)

Determine if the given table should or should not allow its hfiles to be deleted

Parameters:

tableName - name of the table to check

Returns:

true if its store files should be retained, false otherwise

getMonitor

public final HFileArchiveTableMonitor getMonitor()

Returns:

the tracker for which tables should be archived.

create

public static TableHFileArchiveTracker create(org.apache.hadoop.conf.Configuration conf)
                                       throws ZooKeeperConnectionException,
                                              IOException

Create an archive tracker for the passed in server

Parameters:

conf - to read for zookeeper connection information

Returns:

ZooKeeper tracker to monitor for this server if this server should archive hfiles for a given table

Throws:

IOException - If a unexpected exception occurs

ZooKeeperConnectionException - if we can't reach zookeeper

create

private static TableHFileArchiveTracker create(ZKWatcher zkw,
                                               HFileArchiveTableMonitor monitor)

Create an archive tracker with the special passed in table monitor. Should only be used in special cases (e.g. testing)

Parameters:

zkw - Watcher for the ZooKeeper cluster that we should track

monitor - Monitor for which tables need hfile archiving

Returns:

ZooKeeper tracker to monitor for this server if this server should archive hfiles for a given table

getZooKeeperWatcher

public ZKWatcher getZooKeeperWatcher()

stop

public void stop()

Stop this tracker and the passed zookeeper