I use PRTG to monitor a network of about 30 devices and I'm interested in monitoring the routing tables of a few of them. As far I am concerned there is no out of the box sensor provided by PRTG to monitor the entire routing table. Because of this I am using a custom OIDLIB file to add my sensors to each of the devices that I'm interested in. The sensors can be added with no issues. I can attach notification on changes using the "Trigger 'change' notification" setting and I do get the notifications when the routes change.
My issue only occurs when I increase the number of sensors to about >1800. Around 1400 are SNMP Custom string sensors. The problem itself is that every time that I have to restart the server for maintenance, the custom string values don't survive the restart. After a restart the sensors that were supposed to return the IP address of the next hop as specified by the OID, return the string 'OK' instead. This of course has no meaning as far as routes are concerned. Additionally, no notifications are sent even though the last value was an IP address and it now reports the string 'OK'. So it silently fails to report this change.
I have tried to use the Sensor Setting for Keyword Search->Response Not Must Include (Down Status) and provided the string 'OK', but the sensor is never set to down. While the sensor is in this state I used the Paessler SNMP Tester 5.2.3 with the same OID and it returns the IP address as expected so I know the information must be getting to PRTG.
When my installation gets into this unstable state I have to refresh all the SNMP Custom string sensors multiple times so that eventually most of them start reporting the IPs again. On other occasions I've had to delete the sensor and add a new one. Besides time consuming this makes managing the routes extremely painful. I wasn't able to find help in the questions on the knowledge base so I'm posting my issue as a new question.
My questions are: how do I work around this problem so that all of my sensors wake up and start reporting the correct data? If this is a known issue and wont be fixed for a couple of releases I kindly request help to set these SNMP custom string sensors to down when the string 'OK' is reported instead of the route's IP info. This will allow me hook up notifications to these sensors so that I can get them back to report the correct data.
Let me know if you have further questions.