sirius.web.health

Class Cluster

java.lang.Object

sirius.web.health.Cluster

All Implemented Interfaces:

sirius.kernel.timer.EveryMinute, sirius.kernel.timer.TimedTask

public class Cluster
extends Object
implements sirius.kernel.timer.EveryMinute

Manages and monitors the state of a cluster of machines.

Permits to couple a number of machines to a cluster where each member monitors the others. In case or a failure an alert will be triggered. Additionally the cluster state can be visualized using the web interface (/system/state).

Even in a single machine installation, this class will take care of monitoring all metrics and triggering an alert (if possible).

Cluster members are defined via the configuration by listing all HTTP-Endpoints under health.cluster.nodes in the form of http://hostname:port. Each node always defines a priority (health.cluster.priority). The node with the lowest number (which is still functional) is in charge of triggering an alert in case of faulting or unreachable members.

Field Summary

Fields

Modifier and Type	Field and Description
static sirius.kernel.health.Log	LOG

Constructor Summary

Constructors

Constructor and Description
Cluster()

Method Summary

Modifier and Type	Method and Description
NodeInfo	getBestAvailableNode() Returns the best node which is still functional and has the highest priority (lowest number).
sirius.kernel.health.metrics.MetricState	getClusterState() Returns the overall state of the cluster.
List<NodeInfo>	getNodeInfos() Reports infos for all known cluster members.
int	getNodePriority() Returns the priority of this node.
sirius.kernel.health.metrics.MetricState	getNodeState() Returns the determined state of this node.
boolean	isAlarmPresent()
boolean	isBestAvailableNode() Determines if the current node is the best (highest priority, still functional) cluster node.
void	runTimer()

Methods inherited from class java.lang.Object

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

Field Detail

LOG

public static final sirius.kernel.health.Log LOG

Constructor Detail

Cluster

public Cluster()

Method Detail

getNodeInfos

public List<NodeInfo> getNodeInfos()

Reports infos for all known cluster members.

During startup (before the initial communication took place) some information (like the node name) might be missing.

Returns:

a list of all cluster members along with their last known state

getBestAvailableNode

public NodeInfo getBestAvailableNode()

Returns the best node which is still functional and has the highest priority (lowest number).

Returns:

all known data about the best functional cluster node

isBestAvailableNode

public boolean isBestAvailableNode()

Determines if the current node is the best (highest priority, still functional) cluster node.

Returns:

true if the current node is the best node in this cluster

getNodeState

public sirius.kernel.health.metrics.MetricState getNodeState()

Returns the determined state of this node.

Returns:

the state of this node, which is computed every minute

getClusterState

public sirius.kernel.health.metrics.MetricState getClusterState()

Returns the overall state of the cluster.

Returns:

the cluster state, which is the "worst" state among all members

isAlarmPresent

public boolean isAlarmPresent()

runTimer

public void runTimer()
              throws Exception

Specified by:

runTimer in interface sirius.kernel.timer.TimedTask

Throws:

Exception

getNodePriority

public int getNodePriority()

Returns the priority of this node.

Returns:

the priority of this node (lower is better)