On Demand Detection

  • Article

 A really useful feature of Health Explorer in Operations Manager is the ability to force the agent recalculate health of a particular monitor. One the most common uses for this is when you want to confirm that the action that you took actually fixed the problem. If you have implemented your management pack correctly all you need to do is select the unit monitor in health explorer and click on the "Recalculate Health" button.

In order to make this button actually force your monitor to recalculate health state, you must do some extra work in your management pack, or else nothing will happen. This is especially important if you are either creating a new monitor type or creating a script based monitor. Basically when creating a new monitor type or creating a script based monitor, unless you do some extra work "Recalculate Health" will not do much for you and the users of your MP.

Before we dive into the details of what needs to be done, let's first review a standard two state script based monitor. This monitor uses a script to determine the health state of a service (in a shipping MP I would use a different monitor type, but this was but this was convenient for demonstration purposes).

<UnitMonitor ID="Sample.ServiceStatus.Monitor2" Accessibility="Public" Enabled="true" Target="Sample.ServerRole" ParentMonitorID="Health!System.Health.AvailabilityState" Remotable="true" Priority="Normal" TypeID="Windows!Microsoft.Windows.TimedScript.TwoStateMonitorType" ConfirmDelivery="false">

        <Category>Custom</Category>

        <OperationalStates>

          <OperationalState ID="Error" MonitorTypeStateID="Error" HealthState="Error" />

          <OperationalState ID="Success" MonitorTypeStateID="Success" HealthState="Success" />

        </OperationalStates>

        <Configuration>

          <IntervalSeconds>180</IntervalSeconds>

          <SyncTime />

          <ScriptName>GetServiceStatus2.vbs</ScriptName>

          <Arguments>$Target/Host/Property[Type="Windows!Microsoft.Windows.Computer"]/PrincipalName$ dhcp</Arguments>

          <ScriptBody>

            <![CDATA[

                  call Main()

                  Sub Main()

                  Dim oAPI

                  Dim oBag

                  Dim serviceName

                  Dim computerName

                  Dim statusColl

                  Dim serviceStatus

                  computerName = WScript.Arguments(0)

                  serviceName = WScript.Arguments(1)

                                   

                  Set wmi = GetObject("winmgmts:\\" & computerName & "\root\cimv2")

                  wql = "SELECT * FROM Win32_Service WHERE Name='" & serviceName & "'"

                  Set statusColl = wmi.ExecQuery(wql)

                  If statusColl.Count = 1 Then

                    Dim service

                    Set service = statusColl.ItemIndex(0)

                    serviceStatus = service.State

                  End If

                  Set oAPI = CreateObject("MOM.ScriptAPI")

                  Set oBag = oAPI.CreatePropertyBag()

                  oBag.AddValue "ServiceStatus", serviceStatus

                  oAPI.Return(oBag)

                  End Sub

                  ]]>

          </ScriptBody>

          <TimeoutSeconds>120</TimeoutSeconds>

          <ErrorExpression>

            <SimpleExpression>

              <ValueExpression>

                <XPathQuery Type="String">Property[@Name='ServiceStatus']</XPathQuery>

              </ValueExpression>

              <Operator>NotEqual</Operator>

              <ValueExpression>

                <Value Type="String">Running</Value>

       </ValueExpression>

            </SimpleExpression>

          </ErrorExpression>

          <SuccessExpression>

            <SimpleExpression>

              <ValueExpression>

                <XPathQuery Type="String">Property[@Name='ServiceStatus']</XPathQuery>

              </ValueExpression>

              <Operator>Equal</Operator>

              <ValueExpression>

                <Value Type="String">Running</Value>

              </ValueExpression>

            </SimpleExpression>

          </SuccessExpression>

        </Configuration>

      </UnitMonitor>

This monitor looks totally fine and will work properly by running the script and interpreting the results of the script every 180 seconds. The problem is that if you try to recalculate the health state of this monitor using health explorer, nothing will happen. This is because the underlying monitor type which in essence acts as the foundation for this monitor does not implement an "On Demand Detection". The "On Demand Detection" section of the monitor type provides additional information to OpsMgr in order to allow it to recalculate the health state at any point in time rather than waiting for the 180 seconds interval. Now lets look at the steps that need to be taken in order to implement a monitor which will react to a request to recalculate its health state:

1 - Implement a probe which is just responsible for retrieving the particular piece of instrumentation

<ProbeActionModuleType ID="Sample.ServiceStatus.Probe" Accessibility="Public" Batching="false" PassThrough="false">

        <Configuration>

          <xsd:element minOccurs="1" name="TimeoutSeconds" type="xsd:integer" />

          <xsd:element minOccurs="1" name="ComputerPrincipalName" type="xsd:string" />

          <xsd:element minOccurs="1" name="ServiceName" type="xsd:string" />

        </Configuration>

        <ModuleImplementation Isolation="Any">

          <Composite>

            <MemberModules>

              <ProbeAction ID="PassThrough" TypeID="System!System.PassThroughProbe" />

              <ProbeAction ID="Script" TypeID="Windows!Microsoft.Windows.ScriptPropertyBagProbe">

                <ScriptName>GetServiceStatus.vbs</ScriptName>

                <Arguments>$Config/ComputerPrincipalName$ $Config/ServiceName$</Arguments>

                <ScriptBody>

                  <![CDATA[

                  call Main()

               Sub Main()

                  Dim oAPI

                  Dim oBag

                  Dim serviceName

                  Dim computerName

                  Dim statusColl

                  Dim serviceStatus

                  computerName = WScript.Arguments(0)

                  serviceName = WScript.Arguments(1)

                                   

                  Set wmi = GetObject("winmgmts:\\" & computerName & "\root\cimv2")

                  wql = "SELECT * FROM Win32_Service WHERE Name='" & serviceName & "'"

                  Set statusColl = wmi.ExecQuery(wql)

                  If statusColl.Count = 1 Then

                    Dim service

                    Set service = statusColl.ItemIndex(0)

                    serviceStatus = service.State

                  End If

                  Set oAPI = CreateObject("MOM.ScriptAPI")

                  Set oBag = oAPI.CreatePropertyBag()

                  oBag.AddValue "ServiceStatus", serviceStatus

                  oAPI.Return(oBag)

                  End Sub

                  ]]>

                </ScriptBody>

                <TimeoutSeconds>$Config/TimeoutSeconds$</TimeoutSeconds>

              </ProbeAction>

            </MemberModules>

            <Composition>

              <Node ID="Script">

                <Node ID="PassThrough" />

              </Node>

            </Composition>

          </Composite>

        </ModuleImplementation>

        <OutputType>System!System.PropertyBagData</OutputType>

        <TriggerOnly>true</TriggerOnly>

      </ProbeActionModuleType>

2 - Implement a composite data source module which would combine the probe with a scheduler. Its not a must, but I would recommend you do this to simplify the implementation of the monitor type in step 3:

<DataSourceModuleType ID="Sample.ServiceStatus.DataSource" Accessibility="Internal" Batching="false">

        <Configuration>

          <xsd:element minOccurs="1" name="IntervalSeconds" type="xsd:integer" />

          <xsd:element minOccurs="1" name="TimeoutSeconds" type="xsd:integer" />

          <xsd:element minOccurs="1" name="ComputerPrincipalName" type="xsd:string" />

          <xsd:element minOccurs="1" name="ServiceName" type="xsd:string" />

        </Configuration>

        <OverrideableParameters>

          <OverrideableParameter ID="IntervalSeconds" Selector="$Config/IntervalSeconds$" ParameterType="int" />

          <OverrideableParameter ID="TimeoutSeconds" Selector="$Config/TimeoutSeconds$" ParameterType="int" />

        </OverrideableParameters>

        <ModuleImplementation Isolation="Any">

          <Composite>

            <MemberModules>

              <DataSource ID="Scheduler" TypeID="System!System.Discovery.Scheduler">

                <Scheduler>

                  <SimpleReccuringSchedule>

                    <Interval Unit="Seconds">$Config/IntervalSeconds$</Interval>

                  </SimpleReccuringSchedule>

   <ExcludeDates />

                </Scheduler>

              </DataSource>

              <ProbeAction ID="Probe" TypeID="Sample.ServiceStatus.Probe">

                <TimeoutSeconds>$Config/TimeoutSeconds$</TimeoutSeconds>

                <ComputerPrincipalName>$Config/ComputerPrincipalName$</ComputerPrincipalName>

                <ServiceName>$Config/ServiceName$</ServiceName>

              </ProbeAction>

            </MemberModules>

            <Composition>

              <Node ID="Probe">

                <Node ID="Scheduler" />

              </Node>

            </Composition>

          </Composite>

        </ModuleImplementation>

        <OutputType>System!System.PropertyBagData</OutputType>

      </DataSourceModuleType>

3 - Monitor type which contains a section called "OnDemandDetection". The section is highlighted below:

<UnitMonitorType ID="Sample.ServiceStatusMonitorType" Accessibility="Internal">

        <MonitorTypeStates>

          <MonitorTypeState ID="Running" NoDetection="false" />

          <MonitorTypeState ID="NotRunning" NoDetection="false" />

        </MonitorTypeStates>

       <Configuration>

          <xsd:element name="IntervalSeconds" type="xsd:integer" />

          <xsd:element name="TimeoutSeconds" type="xsd:int" />

          <xsd:element name="ComputerPrincipalName" type="xsd:string" />

          <xsd:element name="ServiceName" type="xsd:string" />

        </Configuration>

        <OverrideableParameters>

          <OverrideableParameter ID="IntervalSeconds" Selector="$Config/IntervalSeconds$" ParameterType="int" />

          <OverrideableParameter ID="TimeoutSeconds" Selector="$Config/TimeoutSeconds$" ParameterType="int" />

        </OverrideableParameters>

        <MonitorImplementation>

          <MemberModules>

            <DataSource ID="DS" TypeID="Sample.ServiceStatus.DataSource">

              <IntervalSeconds>$Config/IntervalSeconds$</IntervalSeconds>

              <TimeoutSeconds>$Config/TimeoutSeconds$</TimeoutSeconds>

              <ComputerPrincipalName>$Config/ComputerPrincipalName$</ComputerPrincipalName>

              <ServiceName>$Config/ServiceName$</ServiceName>

            </DataSource>

            <ProbeAction ID="Probe" TypeID="Sample.ServiceStatus.Probe">

              <TimeoutSeconds>$Config/TimeoutSeconds$</TimeoutSeconds>

              <ComputerPrincipalName>$Config/ComputerPrincipalName$</ComputerPrincipalName>

              <ServiceName>$Config/ServiceName$</ServiceName>

            </ProbeAction>

            <ConditionDetection ID="NotRunningCondition" TypeID="System!System.ExpressionFilter">

              <Expression>

                <SimpleExpression>

                  <ValueExpression>

                    <XPathQuery Type="String">Property[@Name='ServiceStatus']</XPathQuery>

                  </ValueExpression>

                  <Operator>NotEqual</Operator>

                  <ValueExpression>

                    <Value Type="String">Running</Value>

                  </ValueExpression>

                </SimpleExpression>

              </Expression>

            </ConditionDetection>

            <ConditionDetection ID="RunningCondition" TypeID="System!System.ExpressionFilter">

              <Expression>

                <SimpleExpression>

                  <ValueExpression>

                    <XPathQuery Type="String">Property[@Name='ServiceStatus']</XPathQuery>

                  </ValueExpression>

                  <Operator>Equal</Operator>

                  <ValueExpression>

                    <Value Type="String">Running</Value>

                  </ValueExpression>

                </SimpleExpression>

              </Expression>

            </ConditionDetection>

          </MemberModules>

          <RegularDetections>

            <RegularDetection MonitorTypeStateID="Running">

              <Node ID="RunningCondition">

                <Node ID="DS" />

              </Node>

            </RegularDetection>

            <RegularDetection MonitorTypeStateID="NotRunning">

              <Node ID="NotRunningCondition">

                <Node ID="DS" />

              </Node>

            </RegularDetection>

          </RegularDetections>

          <OnDemandDetections>

            < OnDemandDetectionMonitorTypeStateID = "Running">

              < NodeID = "RunningCondition">

                < NodeID = "Probe" />

              </ Node >

            </ OnDemandDetection >

            < OnDemandDetectionMonitorTypeStateID = "NotRunning">

              < NodeID = "NotRunningCondition">

                < NodeID = "Probe" />

              </ Node >

            </ OnDemandDetection >

          </OnDemandDetections>

        </MonitorImplementation>

   </UnitMonitorType>

4 - Monitor which uses the monitor type created in step 3.

<UnitMonitor ID="Sample.ServiceStatus.Monitor" Accessibility="Internal" Enabled="true" Target="Sample.ServerRole" ParentMonitorID="Health!System.Health.AvailabilityState" Remotable="true" Priority="Normal" TypeID="Sample.ServiceStatusMonitorType" ConfirmDelivery="false">

        <Category>AvailabilityHealth</Category>

        <AlertSettings AlertMessage="Sample.ServiceStatus.Monitor.AlertMessage">

          <AlertOnState>Error</AlertOnState>

          <AutoResolve>true</AutoResolve>

          <AlertPriority>Normal</AlertPriority>

          <AlertSeverity>MatchMonitorHealth</AlertSeverity>

          <AlertParameters>

            <AlertParameter1>Dhcp</AlertParameter1>

            <AlertParameter2>$Target/Host/Property[Type="Windows!Microsoft.Windows.Computer"]/PrincipalName$</AlertParameter2>

          </AlertParameters>

        </AlertSettings>

        <OperationalStates>

          <OperationalState ID="Running" MonitorTypeStateID="Running" HealthState="Success" />

          <OperationalState ID="NotRunning" MonitorTypeStateID="NotRunning" HealthState="Error" />

        </OperationalStates>

        <Configuration>

          <IntervalSeconds>360</IntervalSeconds>

          <TimeoutSeconds>300</TimeoutSeconds>

        <ComputerPrincipalName>$Target/Host/Property[Type="Windows!Microsoft.Windows.Computer"]/PrincipalName$</ComputerPrincipalName>

          <ServiceName>Dhcp</ServiceName>

        </Configuration>

      </UnitMonitor>

Now the agent will have all the required information to be able to recalculate the health state of the monitor at any given time. The one thing that you should keep in mind is that OnDemandDetection is not possible for every scenario such as event based monitors as the monitor state is driven purely by an occurrence of a particular event and until that event is logged, there is nothing much you can do.

The one area that I would urge you to strongly consider implementing OnDemandDetection is specifically for script based monitors as this is just a matter of refactoring the monitor implementation.

Hope this post provided some additional information on how to build a management pack. If you have questions about OnDemandDetection, please leave me a comment on the blog and I will try to respond to it.

Attached to this post is a sample MP that contains both monitors.

NTServiceBasedDiscoverySample.MP.xml

Comments

  • Anonymous
    June 06, 2008
    PingBack from http://thefurturezblank.com/sampleselector/2008/06/06/sample-selector-on-the-web-13/
  • Anonymous
    July 03, 2008
    The comment has been removed
  • Anonymous
    January 08, 2009
    Can this be used for three-state monitors created in the Console using the wizard? I built an MP using the Windows Service Template to monitor a bunch of custom services and I'm using vbscript similar to the one in your example and I'd like to be able to implement the OnDemandDetection but I'm not sure where I would put the logic as I used the console to build the monitor in the first place.Ideas?Thanks,Bryan
  • Anonymous
    January 08, 2009
    Based on the underlying monitor type, you might not need to implement OnDemandDetection as the monitor type already has it defined. By console, do you mean authoring console or Ops console?
  • Anonymous
    January 13, 2009
    By Console I meant the Ops Console. Sorry for being vague.Thanks.Bryan
  • Anonymous
    January 13, 2009
    Implementing an OnDemandDetection requires writing a new monitor type. This can only be done in the authoring console.
  • Anonymous
    July 04, 2009
    Just downloaded attached MP.is it correct public key?<PublicKeyToken>9396306c2be7fcc4</PublicKeyToken>shouldn't it be "31bf3856ad364e35"?
  • Anonymous
    July 06, 2009
    You are right. The right key to use is the one you pointed out.