Is it possible to configure a notification only if all sensors in the group are down?

Hi Thurmus,

What kind of sensors are you using to monitor the data link at both ends (PING, SNMP...)

I am using the SNMP ones so I could choose the specific port on the switch. This was to help differentiate between a whole switch (PING) and a link. I am happy to use PING, it would just not give me the finite level of detail I was trying to get.

Thanks

Maybe I'm not understanding the situation...I'll admit there are far too many unknowns and it's late (I'm really tired). :0)

Is this a large environment? Are there multiple interconnects/routes between each network segment? Or maybe you have multiple VLANs with multiple ways of notification/communication?

I'm only asking this because if it's a vanilla setup/configuration, I don't see how you could accomplish this goal.

For example: a) Switch01 (1st floor) b) Switch02 (2nd floor) c) SW01 is connected to SW02 via ethernet/fiber d) PRTG is running on a system connected to SW01 e) PRTG checks the sensor associated with the interconnected ports on SW01 and SW02

Situation: 1) Near side interface drops offline 2) PRTG scans the near side interface at its scheduled interval and identifies the error 3) Because the near side interface (and it's now down) connects to the far side, PRTG is unable to successfully check the far side

Comment/Question: 1) Without being able to check the far side interface, PRTG will automatically assume the far side is down as well unless you have established a dependency.

2) If you have PRTG running on systems at both ends, how might one or the other send out a notification if the interconnect is down on one end, the other or both?

3) Why does it only matter that both interfaces are down? Just curious...

We have multiple Fiber rings at the moment which go to a single core. Each end of the ring goes to a different core switch that are linked and spanning tree'd. In each ring the fibre comes into switch port 13 and out switch port 15 of each site. If a Fiber Link to a site dies the traffic will route back to the core from either side of the break. On the ring the fibre going out goes to every second switch and the one coming back goes into the alternate switches. If both the outgoing and incoming fibre are cut I understand it will red everything from that point on. It is a sinle VLAN.

Does this make sense?

There are different people and companies responsible for different parts of this network. If a switch dies it is one groups problem, if a link dies it is another companies problem. I was hoping to be able to differentiate so that the alert could go to the area responsible.

Thanks for your help and interest.

Sensors from the Google Code project

Actually, there are already Custom Sensors available that might suit your needs.

At the Google Code project there are two sensors that could solve your problem.

• PingWarning Use Pingwarning for each device. This sensor will go into a warning state if the device can not be pingged anymore.
• SensorCount Next use SensorCount and set it to count warningsensors from your group downwards. Set the sensor to go "red" if it returns a count of two or more.

A workaround could be the sensor Factory. There you could monitor the sensors of the different devices in one and add the timings. Single problem on the ping sensor which is reporting already an error.

If you create a custom sensor where you replace the return value of the ping by an "return 0" you could do it as followed:

The two sensors for your devices could be monitored in following way

1. 1:AnyLine Channel(4360,0)+Channel(4362,0) You add the timings and in case the result of this factory sensor is "0" you create an error. For the two individual custom sensors you can then configure just to go to warning status.