Total SNMP Confusion

[Rancher]

I am getting ready to rebuild HP Sims from scratch, version 5.4 (currently at 4.2)
I was originally instructed in my initial build by an HP rep, but now, after reading this forum and the instructions for 5.4, I am totally confused on SNMP!

He had me set up both a public (Read) and private (read, create) string on all my field servers and my Sims server. (BTW, I also get two alerts for any event that occurs!)

I would like to get this set up correctly this time so that all SIMS functions work and so I only get one alert per event.

It is my understanding (at least right now) that the private string needs to be on all servers to be managed (even the SIMS server) and the public string only has to be configured on the Sims server?

Or am I even more confused that I thought?

[David Claypool · Accepted solution]

HP SIM only uses a READ community string in its communications with servers. That means that the string has to be present on all of the servers to be managed and you have to tell HP SIM about it.

HP SIM does not use a READ-WRITE or READ-CREATE community string over the wire (all transaction-type operations are conducted using Secure HTTP). However, on the server itself, SNMP READ-WRITE or READ-CREATE strings are used for intra-agent communication (e.g. Insight Agent to Version Control Agent) so a READ-WRITE or READ-CREATE string must be present. HP SIM does not need to know this as it doesn't use it. Since it is never communicated on the wire, it can't be sniffed.

If you use the 'Configure or Repair Agents' tool in HP SIM, it will ensure that HP SIM has proper communications with each server and that intra-agent communications can be completed. It checks to make sure that a READ string known to it is present on the target and if it is not present will add it (any other strings already defined are not changed). If a READ-WRITE or READ-CREATE string is not present it will create a long, random, unique string to establish one.

Finally, it is likely that your dual trap problem is unrelated to this. iLO has a feature to forward traps to HP SIM. This is to allow traps to be sent in the case of a communications failure between the SIM server and the host where presumably iLO might have a different route. Your 2 alerts are probably due to this.

[Rob Buxton]

HPSIM uses the Read Only Community Name in its interactions with the Agents on the Client Servers.
I believe it is the Agents on the Client Server that use the Read Write community name to do things like set thresholds, clear IML Logs etc. No Read / Write activity is sent across the wire from HPSIMN to the client.

So on all of the Client Servers you should have both community names.

In SNMP on the Client Server you also need to configure the Trap Destination using just the Read Only Community name to point back to the HPSIM Server. This is the mechanism that the Agents use to send alerts from failures (e.g. disk) back to HPSIM.

Also check that the HPSIM Server is one of the allowed Hosts in the SNMP config.

[Rancher]

Thank you both so much!
It is all perfectly clear now.

[Jim_]

I have just about 25 servers out of 500 that are having the issue when the Disk Threshold task fails with "The Device is not responding to the SSL request"

The machine SNMP settings are setup just like all the other which match what is suggested in this thread by Rob. The 25 machines are reporting back issues to the SIM console and the VCA is functioning.

Any other ideas on why these 25 are having an issue? I can't seem to find what is different between these 25 and the other 475.

Thanks!

[NJK-Work]

5.4! Did I fall asleep for 2 years and miss version 5.2 and 5.3 :)

Nelson

[Jim_]

Thanks for th humor Nelson :-) Do you have any input on my issue? Thanks,Jim