org.apache.hadoop.ha

Class ZKFCRpcServer

java.lang.Object

org.apache.hadoop.ha.ZKFCRpcServer

All Implemented Interfaces:

ZKFCProtocol

@InterfaceAudience.LimitedPrivate(value="HDFS")
@InterfaceStability.Evolving
public class ZKFCRpcServer
extends Object
implements ZKFCProtocol

Field Summary

Fields inherited from interface org.apache.hadoop.ha.ZKFCProtocol

versionID

Method Summary

Methods

Modifier and Type	Method and Description
void	cedeActive(int millisToCede) Request that this service yield from the active node election for the specified time period.
InetSocketAddress	getAddress()
void	gracefulFailover() Request that this node try to become active through a graceful failover.

Methods inherited from class java.lang.Object

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

Method Detail

getAddress

public InetSocketAddress getAddress()

cedeActive

public void cedeActive(int millisToCede)
                throws IOException,
                       AccessControlException

Description copied from interface: ZKFCProtocol

Request that this service yield from the active node election for the specified time period. If the node is not currently active, it simply prevents any attempts to become active for the specified time period. Otherwise, it first tries to transition the local service to standby state, and then quits the election. If the attempt to transition to standby succeeds, then the ZKFC receiving this RPC will delete its own breadcrumb node in ZooKeeper. Thus, the next node to become active will not run any fencing process. Otherwise, the breadcrumb will be left, such that the next active will fence this node. After the specified time period elapses, the node will attempt to re-join the election, provided that its service is healthy. If the node has previously been instructed to cede active, and is still within the specified time period, the later command's time period will take precedence, resetting the timer. A call to cedeActive which specifies a 0 or negative time period will allow the target node to immediately rejoin the election, so long as it is healthy.

Specified by:

cedeActive in interface ZKFCProtocol

Parameters:

millisToCede - period for which the node should not attempt to become active

Throws:

IOException - if the operation fails

AccessControlException - if the operation is disallowed

gracefulFailover

public void gracefulFailover()
                      throws IOException,
                             AccessControlException

Description copied from interface: ZKFCProtocol

Request that this node try to become active through a graceful failover. If the node is already active, this is a no-op and simply returns success without taking any further action. If the node is not healthy, it will throw an exception indicating that it is not able to become active. If the node is healthy and not active, it will try to initiate a graceful failover to become active, returning only when it has successfully become active. See ZKFailoverController.gracefulFailoverToYou() for the implementation details. If the node fails to successfully coordinate the failover, throws an exception indicating the reason for failure.

Specified by:

gracefulFailover in interface ZKFCProtocol

Throws:

IOException - if graceful failover fails

AccessControlException - if the operation is disallowed