aws_snmp ???

[Guest_imported]

Hi,

We have TNG 2.4.2 installed and monitoring about 30+ servers.
I have not added the dsm ip address in any of the monitored servers aws_snmp.cfg file yet TNG seems to be working fine.

If I do add the dsm ip address in the snmp.cfg file every time the agent polls it local resources (eg. memory, cpu etc) won't a snmp trap
be sent to the DSM ??? (the agent polls vary from about 60 to 300 seconds)

It does not make sense to me to add the extra load on the network to get each trap to the dsm. The dsm polls the agent at specified intervals at 10 minutes or so(I do not know where exactly to verify this), and it should be the agents duty to inform the DSM if any thresholds is breached.

It seems from the forum almost everyone puts the dsm's ip address into the aws_snmp.cfg file, and I cannot find a motivating reason why.

Any comments more than welcome.
faiyazm@edgars.co.za

Regards,
Faiyaz Mayet
South Africa

 

[VDirk]

>> If I do add the dsm ip address in the snmp.cfg file every time the agent polls it local resources (eg. memory, cpu etc) won't a snmp trap be sent to the DSM ???

No, a trap is sent from the agent to the DSM only if there is a state change on the agent:

>> it should be the agents duty to inform the DSM if any thresholds is breached.

Exactly, if the agent has the DSM's IP address in the configuration.
The DSM polling is a heartbeat function (since agent SNMP traps are not 'guaranteed delivered').

 

[JonnoGT]

The DSM will poll all of the agents that it is supposed to be monitoring every 10 mins.
So if you get an event on the agent and have not configured the agent to send traps you will have a 10 min lag before you are notified.

The DSM manages all subnets / machines specified in the gwipflt.dat file (tng\services\config\aws_wvgate). By default this is set to *.

The file on the agent that you need to change is aws_sadmin.cfg not aws_snmp.cfg.
As long as you set your agent thresholds at reasonable values you won't be flooded with messages.

You can also get the "trap translation" messages from the agent that you won't ever get from the DSM polling.

 

[faiyazm]

Hi,

Thanks for the reply.

If I have entered my DSM's Ip address in the server as the SNMP recipent, will I still have to edit aws_sadmin.cfg ??

If the ip address is allready configured in Nt's SNMP settings won't the traps from the agent automatically be sent to the DSM ???

Thank you,
Faiyaz

 

[JonnoGT]

Not 100% sure, but I don't think so.
TNG uses its own version of SNMP, so I think you will have to change the aws_sadmin.cfg file.

You don't actually need SNMP to be running on the agent machine for the agents to work. The only thing you need the SNMP services for is discovery. (But it doesn't hurt to have them).
If, like you have done, you configure MS SNMP you will also get Microsoft messages in the console (raw traps with the 311 Identifier)