Discussion here:

http://www.worldcommunitygrid.org/forums/wcg/viewthread_thread,32488

I think this is how this error occurs:
The error occurs when the computer pauses for a while waiting for some kind of resource, usually related to hardware. The main boinc.exe program regularly asks each science task for some kind of status information. If the science task can't respond in the time allowed because the system has paused, boinc.exe assumes that it has died and terminates it.

A slow or overcommitted hard drive can trigger this condition. CEP2 is very good at overloading your hard drive. Each CEP2 WU contains 16 sub-jobs that run 1 at a time. When each one starts, it creates a huge number of small data files, and the hard drive may not keep up, especially if it is already busy, eg running a virus scan, or starting up another CEP2 WU in a multi-core computer, or having insufficient RAM so that it has to swap tasks out to the pagefile.

CEP2 tasks also generate a huge number of "page faults", where small chunks of data are exchanged between RAM and the pagefile. If only 1-2 CEP2 WUs are running, the HDD activity LED won't show much activity, I think because most of this activity is handled by the operating system's disc cache. However, on multi-core (4 and up) systems that are running CEP2 on most cores, the HDD will become quite busy and the whole machine will take micro-sleeps waiting for the HDD. And then when a new CEP2 job starts making all those little files ...

I had these errors with DDDT2 at one stage, on a machine which had a slow "Green" HDD that I think was defective from new. Although it always passed diagnostics and never gave a hard error, it would occasionally go into super-slow mode and these errors would happen. The HDD activity LED would always be lit brightly and continuously at these times.

Waiting for some other kind of resources such as a network connection might also cause the problem.

If your computer has multiple cores, I would run a mixture of WCG projects and run CEP2 on less than 50% of them. I aim to run CEP2 on 25% of cores. If the errors still occur, try to observe what happens just before an error. Check the behaviour of the HDD activity LED, and check the response when you click on different windows on the screen.

Hope this helps.

Dear PCHooft,
sorry that CEP2 does not seem to be the right fit for your hardware - but the other WCG projects are also great!
Best wishes
Your Harvard CEP team

Start basysitting. The first 100 burbs [heartbeat loss] the task goes back to last checkpoint [a serious time oops on the 3rd job of 16]. At 101 of such heartbeat losses the task is deemed broken and send home. The Result Log will record a heartbeat problem.

--//--

Edit: The not so shy can look in the stderr.txt file of the task-slot while the job is running when they see this happening. (And on foresight, the developers have hidden the message log, for the over concerned parents... the message tab was removed and in place a Notices tab was put that will generate a pop up and a blinking red button if user intervention is required. To quote the developer yesterday: "The event log is for developers and testers. Information for normal users goes in the other parts of the GUI. If we're not showing something important, the solution is to figure out where to show it."). Zero status comes with the "If this happens frequently..." i.e. if occasionally... move on".