ZenOss and Cacti Issues

First off hello and thanks for allowing me to join your site.

I am working on an issue that I haven't been able to resolve. Thought I would reach out to see if anyone has any recommendations.

Currently have a new environment that we are setting up for a new environment that is based on HP Blade hardware running Unix VM's for the application. We have setup 2 Admin Server VM's in two different data centers. We recently installed Cacti Hosts on two of them to monitor traffic and create reporting data for TPS and performance tuning/capacity planning, (currently use Cacti on our legacy system so management wants that used in the new environment.) Problems I am seeing are now the two servers that are running Cacti are constantly showing "Unable to read process on device" and Network connection errors. Inside the ZenOss site it shows that the nodes are bouncing up/down but the box is stable and functioning properly. The issue can be reproduced via command line. It appears to be a networking problem on the TOS01 node in both data centers. When I manually query it it starts returning data, then slows down then times out. Subsequent retries showed 100% timeout, connect with timeout

I surmise the reason for the multiple emails is as follows:

• Zenoss connects successfully to the TOS node and pulls snmp polling data.
• Zenoss then times out attempting to poll the TOS this triggers an SNMP down alarm and email is sent out as its seen as the 1st occurrence of the alarm.
• Zenoss then re-connects to the TOS and pulls snmp polling data this clears the SNMP down alarm.
• This sequence repeats.(Alarm, Email, Successful Poll / Clear)

This hypothesis is backed up by the current count in Zenoss showing only 2 failures. That Cacti and ZenOSS are competing for the same SNMPD resource(s).

System VM O/S: Linux 2.6.32-642.el6.x86_64

ZenOSS Version: ZenOSS5

Anyone got any ideas on how to get the two to play nicely