This may be a bug. Zenoss wasn't able to reproduce, but if you can add / re-open the ticket and work with them, they may be able to work out what's causing this.

See

http://dev.zenoss.org/trac/ticket/7870

--

James Pulver

Information Technology Area Supervisor

LEPP Computer Group

Cornell University

sec:

I would reopen the ticket if you have time even if the problem is resolved after a custom compile. I'm interested to see if it reoccurs or not in your compiled version. Keep us posted.

Best,
--Shane W. Scott (Hackman238)

i am facing exactly the same issue. We are on the way to migrate from a productive Zenoss Core v 2.4.5 to current v3.2.1.

Monitoring Oracle Solaris 10 Servers (Global Zones & Non Global Zones/Containers) on both Zenoss with exactly the same pattern whereas the following processes are modeled on both but alerting as down on v3.2.1 only:

zCountProcs set to false on both (tried also true).

- pattern=ora_dbw - only one process (ora_dbw1) is up wheras all others according the pattern (ora_dbw0, 2 - n) are down.

- pattern=snmpd - says down which is not true

- pattern=^sched (Despite zsched processes are discovered?) - snmpd - says down which is not true

I think v3.2.1 does not deal properly with OS Processes as soon a pattern matches more than one process (e.g. On a global zone there are also all the container processes visible.

Just found a difference on IgnoreParameters, maybe due to trying several settings to fix this.Also after makeing it equal there is no change in alerting behaviour.

Another strange behaviour is, when i change globally the name/pattern of a process, on the device the process class does not reflect this. (snmp >> snmpd, still process class is snmp)