kill a switch with Cacti?

[knobdy]

Anyone ever heard of Cacti killing a switch?

We had two go down last week. They're pretty much configured the exact same and have the same hardware (C3500XL-C3H2S-M). The only log messages we saw for them all related to snmp - and they had run out of memory.

We've now upgraded the software running on them and have re-enabled them in Cacti. Just wondering if anyone else had heard of such a thing being caused by Cacti or what else we might look at...

[JJX]

I am already monitoring many cisco switches without any problem.
Can it be a problem(=bug?) with snmp?

[knobdy]

I suppose. The only thing new is that I changed to using the cisco router templates (found on this site) to monitor them. But I'm doing the same for all of our switches and these are the only two having problems. Opening a ticket up with Cisco (I don't think this has anything to do with Cacti - but I wanted to eliminate it as a possible cause).

[BSOD2600]

The only log messages we saw for them all related to snmp - and they had run out of memory.

Sounds like a bug with snmp in the IOS.

[JJX]

............. (I don't think this has anything to do with Cacti - but I wanted to eliminate it as a possible cause).

The template works 100%
Check the IOS version and possible bugs

Is possible to pm me with the case id so i can watch it?
thx

[gandalf]

Well, watching some of those "weird" snmp agents out there, this does not sound that strange to me. Of course, even an snmp agent may contain a bug and eat up all your memory. Or, if queried for LOTS of OIDs, the agent may consume the whole cpu ...
All of them were already encountered in the wild ...
Reinhard

[Jeff_Smith]

From a reliability I find it kind of distrubing that a external found (in this case SNMP requests from Cacti) could lock up your switch. I would get after Cisco if I where you. Luckily this was unintentional but imagine an unhappy employee do this on purpose.