com.yahoo.search.cluster

Class BaseNodeMonitor<T>

java.lang.Object

com.yahoo.search.cluster.BaseNodeMonitor<T>

Direct Known Subclasses:

TrafficNodeMonitor

public abstract class BaseNodeMonitor<T>
extends Object

A node monitor is responsible for maintaining the state of a monitored node. It has the following properties:

• A node is taken out of operation if it fails
• A node is put back in operation when it responds correctly again responseAfterFailLimit times unless it has failed failQuarantineLimit. In the latter case it won't be put into operation again before that time period has expired

Author:

bratseth

Field Summary

Fields

Modifier and Type	Field and Description
protected MonitorConfiguration	configuration The configuration of this monitor
protected long	failedAt The last time this node failed, in ms
private boolean	internal Is the node we monitor part of an internal Vespa cluster or not
protected boolean	isQuarantined Whether this node is quarantined for unstability
protected boolean	isWorking
protected static Logger	log
protected T	node The object representing the monitored node
protected long	respondedAt The last time this node responded (failed or succeeded), in ms
protected long	succeededAt The last time this node responded successfully

Constructor Summary

Constructors

Constructor and Description
BaseNodeMonitor(boolean internal)

Method Summary

Modifier and Type	Method and Description
abstract void	failed(ErrorMessage error) Called when this node fails.
T	getNode()
boolean	isIdle()
boolean	isInternal() Returns whether or not this is monitoring an internal node.
boolean	isQuarantined()
boolean	isWorking() Returns whether this node is currently in a state suitable for receiving traffic.
protected long	now()
abstract void	responded() Called when a response is received from this node.
protected abstract void	setWorking(boolean working, String explanation) Thread-safely changes the state of this node if required

Methods inherited from class java.lang.Object

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

Field Detail

log

protected static Logger log

node

protected T node

The object representing the monitored node

isWorking

protected boolean isWorking

isQuarantined

protected boolean isQuarantined

Whether this node is quarantined for unstability

failedAt

protected long failedAt

The last time this node failed, in ms

respondedAt

protected long respondedAt

The last time this node responded (failed or succeeded), in ms

succeededAt

protected long succeededAt

The last time this node responded successfully

configuration

protected MonitorConfiguration configuration

The configuration of this monitor

internal

private boolean internal

Is the node we monitor part of an internal Vespa cluster or not

Constructor Detail

BaseNodeMonitor

public BaseNodeMonitor(boolean internal)

Method Detail

getNode

public T getNode()

isWorking

public boolean isWorking()

Returns whether this node is currently in a state suitable for receiving traffic. As far as we know, that is

isQuarantined

public boolean isQuarantined()

failed

public abstract void failed(ErrorMessage error)

Called when this node fails.

Parameters:

error - a description of the error

responded

public abstract void responded()

Called when a response is received from this node. If the node was quarantined and it has been in that state for more than QuarantineTime milliseconds, it is taken out of quarantine. if it is not in quarantine but is not working, it may be set to working if this method is called at least responseAfterFailLimit times

isIdle

public boolean isIdle()

now

protected long now()

setWorking

protected abstract void setWorking(boolean working,
                                   String explanation)

Thread-safely changes the state of this node if required

isInternal

public boolean isInternal()

Returns whether or not this is monitoring an internal node. Default is false.