Process monitoring not generating events

Community Support
#1

We are using OpenNMS 26.1. We decided to try process monitoring using either the HostResourceSwRunMonitor or the PrTableMonitor. I’ve configured both under the “example1” package. I attempted to define a process service for crond using the send-event syntax for registering a new service in one case, but it didn’t seem to work.

I am able to connect to the Karaf shell and run opennms:poll to get the status of my processes. I can detect state changes and the outputs of these polls show up in polled.log.

I have checked and rechecked the polling intervals in pollard-configuration.xml and syntax seems to match examples.

Yet… I get no events, and I have no associated process services. I should be polling every 5 minutes as well for processes, and the pollen.log (in debug mode) shows no such polling for processes I’ve defined.

Any help is appreciated. I seem to think I need a service defined, but my attempt to do that did nothing. A link to a process monitoring and data collection wiki page suggested more set up info there, but the log says the page was deleted in December.

Thanks for any help! I have spent much time on this. We are on an isolated network is copy and pasting is difficult, as is running any patches.

Thanks again!
Dan K

#2

Are you getting other SNMP metrics collected? These process monitors pull status during SNMP collection and require that to work.

#3

Yes, SNMP metrics collection is working. We routinely get Memory (swap) alerts, disk space threshold exceeded alerts, and the like, generated via the SNMP collection.

I am just wondering if we are missing something? I only have the process services described in pollerd-configuration.xml, as the example in the admin guide (copied below). There was a note in another wiki page, referencing a deleted article, about needing to define something in datacollection.xml? I don’t have anything in there, but maybe that is deprecated? And what about actually defining a service on a node, ie. Process-crond as a service on a node myhost.example.com, such that it shows up in the Availability bar graph?

THank you!

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

<monitor service="Process-httpd" class-name="org.opennms.netmgt.poller.monitors.HostResourceSwRunMonitor"/>
#4

Just as a verification, see if you can walk the hrSWRunTable from your OpenNMS server.

snmpwalk -v 2c -c community host HOST-RESOURCES-MIB::hrSWRunName | grep httpd
(replacing community and host as appropriate)

#5

We can, including the statuses, etc. polling from Karaf works with the OpenNMS:poll command. There is no other config or collection needed for this to work? No definition of services on nodes?

Thank you!

#6

opennms:poll in Karaf should trigger a poll based on the corresponding service in poller-configuration.xml. For this, I think you may want to verify you can opennms:collect in Karaf, as that is the process that collects metrics for storage and the graphs. “Polling” just returns an up/down status and doesn’t necessarily persist data for the graphs on its own.