5

We are trying to monitor our servers mainly with SNMP. Due to performance reasons we are changing this from single requests to snmp-bulk-requests (as allowed in SNMP v2c).

In theory (at least to my knowledge) it should be possible to request several branches/values in a single bulk-request, so the number of tcp-sessions needed for this will be minimal.

We are currently using CentOS 5 for this - the high-level-command in question is snmpbulkwalk - but this only seems to be able to request a single branch (i.e. subtree) at a time.

Am I wrong in theory, or is there a high-level-command out there that can do what I want?

Improve this question
0

2 Answers 2

Reset to default
2
+100

The command snmpbulkget does allow you to specify arbitrary, non-contiguous OID requests. Getting the non-repeaters and max-repetitions right may require some experimenting.

There is a good example here: http://docstore.mik.ua/orelly/networking_2ndEd/snmp/ch02_06.htm

Improve this answer
0

You're right. snmpbulkwalk is a lot more network efficient. However, SNMP does not usually use TCP; it uses UDP.

SNMP bulk requests are not in the SNMPv1 specs, so make sure your SNMP agents you're requesting from is at least SNMPv2c or above.

If you want, you could wrap the snmpbulkwalk in a script to help with your batch bulk request job.

If you're wanting to monitor a lot of information, you might want to look into using RRDTool or a full-blown monitoring system, like Nagios.

Improve this answer
1
  • The agents are v2c capable (mostly Linux). It is actually Nagios that is using these single snmp-requests. So the idea is to make a single big bulk-request and let the Nagios-"Monitors" evaluate the results of that single request (instead of having each monitor issue its own request).
    – Nils
    Aug 26, 2012 at 20:48

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .