Garbage Collection Time

Microsoft.JEE.PerformanceHealth.GarbageCollectorTimeMonitor (UnitMonitor)

The performance monitor for the garbage collection time of an application server.

Knowledge Base article:

Summary

The garbage collection time monitor tracks the time taken by the garbage collector to perform garbage collections on the Java Virtual Machine associated with the application server.

Causes

An unhealthy state indicates that the garbage collector has exceeded the monitored threshold configured for the application server.

There are two distinct causes for the garbage collection time to exceed the threshold:

Insufficient resources (memory) that cause the garbage collectors to activate frequently.
Too many resources (memory) allocated to the Java Virtual Machine. This causes the garbage collector to activate less frequently; however, the time taken for a single collection exceeds the threshold.

Resolutions

If the cause is due to insufficient resources, then allocating more resources to the Java Virtual Machine will alleviate the problem.

If the cause is due to too many resources, then removing resources from the Java Virtual Machine will alleviate the problem.

Element properties:

Target

Microsoft.JEE.GarbageCollector

Parent Monitor

System.Health.PerformanceState

Category

PerformanceHealth

Enabled

False

Alert Generate

True

Alert Severity

MatchMonitorHealth

Alert Priority

Normal

Alert Auto Resolve

True

Monitor Type

Microsoft.JEE.ThresholdDeltaUnitMonitor

Remotable

True

Accessibility

Public

Alert Message

Garbage collector time monitor exceeded

The garbage collector time monitor exceeded its threshold for application server ID: {0}, Object Name: {1}, Counter Name: {2}, Instance Name: {3}

RunAs

Default

Source Code:

<UnitMonitor ID="Microsoft.JEE.PerformanceHealth.GarbageCollectorTimeMonitor" Accessibility="Public" Enabled="false" Target="Microsoft.JEE.GarbageCollector" ParentMonitorID="Health!System.Health.PerformanceState" Remotable="true" Priority="Normal" TypeID="Microsoft.JEE.ThresholdDeltaUnitMonitor" ConfirmDelivery="true">

  <Category>PerformanceHealth</Category>

  <AlertSettings AlertMessage="Microsoft.JEE.PerformanceHealth.GarbageCollectorTimeMonitor_AlertMessageResourceID">

    <AlertOnState>Warning</AlertOnState>

    <AutoResolve>true</AutoResolve>

    <AlertPriority>Normal</AlertPriority>

    <AlertSeverity>MatchMonitorHealth</AlertSeverity>

    <AlertParameters>

      <!-- The alert description should include key properties of:   -->

      <!-- * the deep monitored application server                   -->

      <!-- * the performance counter object name                     -->

      <!-- * counter name                                            -->

      <!-- * instance name                                           -->

      <AlertParameter1>$Target/Host/Property[Type="Microsoft.JEE.ApplicationServer.Monitored.Instance"]/Id$</AlertParameter1>

      <AlertParameter2>Garbage Collector</AlertParameter2>

      <AlertParameter3>JVM Garbage Collector Collection Time Change Rate</AlertParameter3>

      <AlertParameter4>$Target/Property[Type="Microsoft.JEE.GarbageCollector"]/GarbageCollectorName$</AlertParameter4>

    </AlertParameters>

  </AlertSettings>

  <OperationalStates>

    <OperationalState ID="Microsoft.JEE.PerformanceHealth.GC.Time.UnderThreshold" MonitorTypeStateID="UnderThreshold" HealthState="Success"/>

    <OperationalState ID="Microsoft.JEE.PerformanceHealth.GC.Time.OverThreshold" MonitorTypeStateID="OverThreshold" HealthState="Warning"/>

  </OperationalStates>

  <Configuration>

    <ComputerName>$Target/Host/Property[Type="Microsoft.JEE.ApplicationServer.Monitored.Instance"]/HostName$</ComputerName>

    <Protocol>$Target/Host/Property[Type="Microsoft.JEE.ApplicationServer.Monitored.Instance"]/Protocol$</Protocol>

    <Port>$Target/Host/Property[Type="Microsoft.JEE.ApplicationServer.Monitored.Instance"]/Port$</Port>

    <BaseURL/>

    <URLSuffix/>

    <IntervalSeconds>900</IntervalSeconds>

    <TimeoutSeconds>300</TimeoutSeconds>

    <!-- Need to find the GCCollectionTime that has a sibling element      -->

    <!-- called GCName with a value of (for example) "PS Scavenge"         -->

    <!-- XPath should look like:                                           -->

    <!-- /Stats/GC/Properties/GCCollectionTime[../GCName = "PS Scavenge"]  -->

    <MetricName>/Stats/GC/Properties/GCCollectionTime[../GCName = "$Target/Property[Type="Microsoft.JEE.GarbageCollector"]/GarbageCollectorName$"]$</MetricName>

    <!-- The spec says that a value of 1000ms per GC is the threshold        -->

    <!-- as the monitor cannot calculate the values a value of 5000ms        -->

    <!-- has been chosen for the 300 second monitor frequency                -->

    <!-- This equates to roughly 5 GC's per 300 seconds of under 1000ms each -->

    <Threshold>5000</Threshold>

    <SampleCount>2</SampleCount>

  </Configuration>

</UnitMonitor>