Checking past cases was no help so here are a few things from a general perspective.
If the problem started only after upgrading to 2.40 try going back to 2.33 or whatever your prior version was.
Logs are usually not much help for something like this but both the probe log and OS messages log should be checked.
Another option is to disable all monitoring except for one metric and keep adding until the problem happens.
Or it could be that the probe is having a load problem, processing more than it can handle.
For normal operation keep the log level to 0 or 1 which will help reduce the load.
It is a java probe so increasing the memory allocation may help.