Hi Ron,

We always ping first (unless you disable ping) because that way the poller can skip down devices faster than waiting for SNMP (retries +1) * timeouts.

Do I understand your question is that you would like to know if the device is not down, but SNMP on it is?

Have I got this for you:

save image

+

save image

This will give you different alerts based on whether ping or snmp is the issue.

Tom

On 2015-12-03 14:44, Ron Culler wrote:

I have noticed that I see down notices in Observium often when the snmp check fails (I have only seen this when checking servers (Windows/Linux). The device is up and a ping/icmp check would see this but because the snmp check had a problem I get a down indication. Sometimes it has to deal with the snmp service on the host being checked hanging or crashing, other times I have seen load or congestion cause it.

Is there an option or would it make sense to add a feature that would check snmp first and in the event of an error fall back and do a ping/icmp check to see if the device is up or not? It would be beneficial to see if the server is up and get a notice about a snmp check failure. I could also see a check flow that would check snmp first, if it errored it would check with a ping, if the ping returns wait for the next snmp check before sending the notice or marking the device down.

Ron Culler

---

observium mailing list observium@observium.org http://postman.memetic.org/cgi-bin/mailman/listinfo/observium