From OpenNMS

Contents 1 Description 2 Prerequisites 3 HostResourceSwRunMonitor Parameters 3.1 Monitor-specific required parameters 3.2 Monitor-specific optional parameters 3.3 Parameters common to all monitors 4 Example config 4.1 capsd-configuration.xml 4.2 poller-configuration.xml 5 Version History/Availability

Description

This poller monitor's job is to monitor the running state of one or more processes. It does this via SNMP by inspecting the hrSwRunTable of the HOST-RESOURCES-MIB, matching values of hrSwRunName against a configured string, and for rows that match comparing the value of hrSwRunState against an optional numeric value.

Prerequisites

This monitor uses SNMP to accomplish its work. Therefore systems against which it is to be used must have an SNMP agent installed and configured. If OpenNMS is able to discover and collect data from the SNMP service on a given node, then that first requirement has almost certainly been met. Furthermore, the SNMP agent on the system must support the HOST-RESOURCES-MIB; most modern SNMP agents (including most distributions of the Net-SNMP agent and the SNMP service that ships with Microsoft Windows) support this MIB. Out-of-box support for HOST-RESOURCES-MIB among commercial Unix operating systems may be somewhat spotty.

HostResourceSwRunMonitor Parameters

Monitor-specific required parameters

service-name 

The name of the process to be monitored. This parameter's value is case-sensitive and is evaluated as an exact match.

Monitor-specific optional parameters

match-all 

If the process name appears multiple times in the hrSwRunTable, and this parameter is set to true, then all instances of the named process must match the value specified for run-level.

run-level 

The maximum allowable value of hrSWRunStatus among {running(1), runnable(2) = waiting for resource, notRunnable(3) = loaded but waiting for event, invalid(4) = not loaded}. Defaults to 2 (runnable).

service-name-oid 

The numeric object identifier (OID) from which process names are queried. Defaults to hrSwRunName == .1.3.6.1.2.1.25.4.2.1.2 and should never be changed under normal circumstances.

service-status-oid 

The numeric object identifier (OID) from which run status is queried. Defaults to hrSwRunStatus = .1.3.6.1.2.1.25.4.2.1.7 and should never be changed under normal circumstances.

Parameters common to all monitors

ds-name

rrd-repository

port

timeout

retry

Example config

As an example, let's check for a process called httpd using this monitor.

capsd-configuration.xml

Currently the SnmpPlugin lacks a way to discover the presence of processes for polling with this monitor . It's possible, however, to use the LoopPlugin to force the appearance of such a service on one or more interfaces. For instance, to make the service described above appear on all interfaces with IP addresses in the range 192.168.42.50 through 192.168.42.69, and on 192.168.42.42:

<protocol-plugin protocol="Process_httpd"
                 class-name="org.opennms.netmgt.capsd.plugins.LoopPlugin"
                 scan="off" user-defined="false">
  <protocol-configuration scan="enable" user-defined="false">
    <range begin="192.168.42.50" end="192.168.42.69" />
    <specific>192.168.42.42</specific>
  </protocol-configuration>
</protocol-plugin>

As an alternative to Capsd, provisioning groups may be used. In this case the service will still need to be defined in capsd-configuration.xml, but it need not be forced onto any interfaces.

With this configuration every httpd process on all machines will be detected. To limit the detection to specific nodes, you can use the protocol-configuration like the example above

  <protocol-configuration scan="enable" user-defined="false">
    <range begin="192.168.42.50" end="192.168.42.69" />
    <specific>192.168.42.42</specific>
  </protocol-configuration>

In OpenNMS version 1.6.4 and later is the HostResourceSwRunPlugin available. It allows to detect running processes. The configuration should look like this

<protocol-plugin protocol="Process_httpd"
                 class-name="org.opennms.netmgt.capsd.plugins.HostResourceSwRunPlugin"
                 scan="on" user-defined="false">
  <property key="timeout" value="2000" />
  <property key="retry" value="1" />
  <property key="service-name" value="httpd" />
</protocol-plugin>

poller-configuration.xml

To test that at least one process called httpd in a running or runnable state, the following service definition will work:

<service name="Process_httpd" interval="300000" user-defined="false" status="on">
  <parameter key="retry" value="1"/>
  <parameter key="timeout" value="3000"/>
  <parameter key="service-name" value="httpd"/>
</service>

For a slightly more advanced example, let's test that at least one process called httpd exists, and that all processes with that name are in a valid state, i.e. have a run-level no higher than notRunnable(3):

<service name="Process_httpd" interval="300000" user-defined="false" status="on">
  <parameter key="retry" value="1"/>
  <parameter key="timeout" value="3000"/>
  <parameter key="service-name" value="httpd"/>
  <parameter key="run-level" value="3"/>
  <parameter key="match-all" value="true"/>
</service>

Important Notes:

• The name attribute of the service in poller-configuration.xml needs to match the protocol attribute of the protocol-plugin in capsd-configuration.xml.
• You'll also need a line to map the new service to a monitor class (see at the end of the file):

<monitor service="Process_httpd"         class-name="org.opennms.netmgt.poller.monitors.HostResourceSwRunMonitor"/>

Version History/Availability

• This feature was added in version 1.5.94