Throwing vmodl.fault.RequestCanceled without actually having cancelled a request

I'm running VMware Server 2.0.0, Build 122956, and Vix 1.6.2, Build 127388, on an Ubuntu 8.04.03-LTS system. The setup works reasonably well most of the time - I've been able to do VM management via the Perl portion of the API, and generally speaking, it's good to go.

The reason I'm here is that I'm having an intermittent failure of the system that makes no sense (as best I can tell). The system is primarily used to manage a pair of virtual machines that I'm using to run a virus testing sandbox, with my Perl script running a process that boils down to:

• Grab infected file, copy it over to clean virtual machine A

• Start tcpdump on the host OS (which requires me to fork off a child process so it doesn't block)

• Run the infected file on virtual machine A (clean virtual machine B is just sitting there so that it can be infected, and I can capture the traffic going over the network)

• Wait 5 minutes

• Kill the tcpdump process (no further fork, just a syscall)

• Revert both virtual machines to their clean, pre-infected state

• Loop back to the start with another file.

It works well for about 2-3 days, before my Perl script hangs during the process of reverting the second virtual machine. At that point, my hostd.log shows the following:

Activation : Invoke done on

Throw vmodl.fault.RequestCanceled

Result:

(vmodl.fault.RequestCanceled) {

   dynamicType = ,

   msg = ""

}

Activation : Invoke done on

Throw vmodl.fault.RequestCanceled

Result:

(vmodl.fault.RequestCanceled) {

   dynamicType = ,

   msg = ""

}

Activation : Invoke done on

Throw vmodl.fault.RequestCanceled

Result:

(vmodl.fault.RequestCanceled) {

   dynamicType = ,

   msg = ""

}

RecordOp REMOVE: sessionList, ha-sessionmgr

Event 3653 : User akirk logged out

RecordOp ASSIGN: latestEvent, ha-eventmgr

Obviously, I'm not manually issuing a cancellation request of any kind, since the goal is to just let my script run as long as it has files to process (which is essentially infinitely, given the amount of data I'm trying to look at). Is there any other reason such a request might be issued? Do I need to do a host disconnect and reconnect on each iteration of my loop? Or can VMware Server and/or VIX just not handle a long-running loop like that well?