We have a massive device configured with a ton of NetFlow sensors for monitoring a site's individual VLANs independently and have recently run into a processor utilization issue with the probe server it is currently configured for, necessitating a move to a new server we've allocated just for this purpose. Upon moving the device to the new probe at the site, our entire PRTG installation began having intermittent response to all probes in our entire organization AND to sensors localized on the core server. Even after allowing things to stabilize over a 4 hour period, the intermittent sensor response across the board did not improve and the NetFlow sensors on the moved device never came up.
In order to attempt to determine the cause of this issue, all NetFlow sensors on the moved device were paused except two of them. After doing this, all intermittent response issues cleared up over the following 10-20 min but the two unpaused sensors first went into 'Down' state indicating that the socket could not be bound because the address and port were already in use.
After verifying 100% certain that the port was open and ensuring that the IP was not a duplicate, I then decided to clone one of the unpaused sensors and also create a brand new sensor with the same data on the existing and the cloned sensor to see if any of them would respond. After waiting about 5 min, the only sensor that would respond was the brand new sensor that was created so it seems that the NetFlow sensors somehow link to the probe they were created on. That doesn't make sense to me and I could not find this behavior documented anywhere so could this be a build issue? I tried an older build we had on a separate server and tested a similar circumstance and ran into the same problem.
If this is not a build issue, can you please let me know how I can migrate these sensors to the new probe? Is there a way I can export the data, modify it, and then import it? The amount of information we have configured on each requires that we duplicate settings exactly and according to a time estimate I put together based on performing this for 5 of the sensors it will take between 4 and 6 hours to recreate every one of these sensors which is time I don't have...
Add comment