10-18-2011 10:04 PM
This is an odd one and for the life of me, I don't know how to solve it.
We have a linux machine (redhat) that has 5 interfaces:
eth2 = 10.16.122.31
bond0 (has eth0 and eth1) = 10.0.0.1
bond1 (has eth4 and eth5) = 10.0.32.1
When I add a node with HPOV NNM v7.50, I use 10.16.122.31 as an IP address. It then adds the node, but later on refers to it as '10.0.0.1'. When alarms are received, I can see that HPOV runs a whole bunch of GETNEXT requests to poll the device. However, if I right click on the node and select 'alarms' I get nothing besides:
system name changed (was 10.16.122.31) and the source is 10.0.0.1
If I change the filter (and add 10.16.122.31), then the alarms are shown in the tab. But I think HPOV disregards any alarm sent from that device. I tried to change the Preferred SNMP address, to no avail.
The agent is net-snmp 5.3.
Can you please help? I'm baffled. If I send a trap from the agent on a different app, the trap is received correctly and looks perfectly fine (coming from 10.16.122.31). Thanks!
10-19-2011 05:36 AM - edited 10-19-2011 05:38 AM
Yes NNM 7.5x can be rather awkward in this respect. In your case it chooses the lowest IP address as preferred SNMP management address. A netmon lrf option (-k adjustNodeSnmpAddr=false) was added in NNM 7.53 that should stop netmon from switching preferred management addresses.
But you can also try the following:
- Temporarily unmanage all the interfaces for the node except the interface that is to be set as the SNMP Address.
- Issue a demand poll for the node.
- Check the object properties of the node for SNMP address.
- Re-manage the rest of the node's interfaces.