I have seen a similiar problem since v2 beta 12. Pings fail, sometimes takes hours before they declare the device up again, yet all other probes are fine (SNMP,etc). We have seen this on various different devices (APC devices, Netopia, Trango, Mikrotik RBs, etc). I had to turn off notification for our network until we figure out what to do. Pings from the NMS CLI work fine....no packet loss. Only thing I can think of is that Dude is dropping the response messages due to buffering or the like? Maybe it can't handle x number of device responses at one time? I have our pings set to a one minute interval.
How many devices do you have in your network?
I should add that our network is L3 routed and we see the issue on devices in all subnets.