I've seen it before when I hadn't made an exception in an antivirus solution for the BOINC/WCG data directory - i.e., a file sharing/locking conflict.

Then again, I've also seen it when I had failing RAID sets (because the drivers and/or hardware weren't up to snuff...not to mention names, but I still marvell at that particular failure). I'd check system/event logs first.

A very busy system could cause this too [and CEP2 will do that on underpowered systems]. One more reason why this science is opt-in and 1 task at the time.

There's a Start Here topic on this btw. Look in the stickied index [over all forums].

--//--

Not sure if this fits into this thread but one (out of 4 that I selected to run CEP2) of my hosts terminated all CEP2 tasks thrown at it within a matter of minutes:

Result Name: E206033_ 543_ C.22.C16H10OS2SeSi2.01441855.0.set1d06_ 0--
<core_client_version>6.10.58</core_client_version>
<![CDATA[
<message>
- exit code 195 (0xc3)
</message>
<stderr_txt>
INFO: No state to restore. Start from the beginning.
[13:06:46] Number of jobs = 16
[13:06:46] Starting job 0,CPU time has been restored to 0.000000.
Application exited with RC = 0x9
[13:09:40] Finished Job #0
13:09:40 (3872): called boinc_finish

</stderr_txt>
]]>

What gives?
This is a machine that is currently sitting idle, and should not be "underpowered" either

		Starting BOINC client version 6.10.58 for windows_x86_64
		log flags: file_xfer, sched_ops, task
		Libraries: libcurl/7.19.7 OpenSSL/0.9.8l zlib/1.2.3
		Running as a daemon
		Data directory: C:\ProgramData\BOINC
		Running under account boinc_master
		Processor: 2 GenuineIntel          Intel(R) Pentium(R) CPU B940 @ 2.00GHz [Family 6 Model 42 Stepping 7]
		Processor: 256.00 KB cache
		Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 
clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 cx16 sse4_1 sse4_2 nx lm tm2 popcnt pbe
		OS: Microsoft Windows 7: Home Premium x64 Edition, Service Pack 1, (06.01.7601.00)
		Memory: 3.95 GB physical, 7.90 GB virtual
		Disk: 455.34 GB total, 377.40 GB free
		Local time is UTC -8 hours
		No usable GPUs found
	World Community Grid	URL http://www.worldcommunitygrid.org/; Computer ID 1757947; resource share 100
	World Community Grid	General prefs: from World Community Grid (last modified 14-Feb-2012 18:58:00)
	World Community Grid	Computer location: work
		General prefs: using separate prefs for work
		Preferences:
		   max memory usage when active: 2021.93MB
		   max memory usage when idle: 3032.89MB
		   max disk usage: 10.00GB
		   don't use GPU while active
		   suspend work if non-BOINC CPU load exceeds 50 %
		   

GFAM WUs on this machine work just fine, so do the CEP2 WUs on the other 3 hosts that have that project in their selection...

Ralf

Well, after 5 WUs in a row failed on that box, I switched the device profile to one that doesn't include CEP2. Can give it a try again once weekend comes along, hopefully...

Maybe didn't search right, but didn't see anything that fit, that's why I mentioned it.
3 other hosts enabled for CEP2 at the same time work just fine, with the only real difference obvious to me is that those are running 32bit Windows XP Pro/Windows Server 2003, while the "problematic" one is Windows 7 64bit...

Ralf

From an off-line discussion, an option is looked for to automate the maximum number of ''intense'' jobs. Something that members can configure in their device profile and the client would of course have as a feature... the next client after the one after the following one, who knows. Would require both server and client side coding change. Then the client could have as many CEP2 in the buffer, but if you have set e.g. 2 intense concurrent on a quad, the client would only let 2 run at the time. Of course you'd need other work too, else these remaining cores would idle.

Something along these lines, then everyone can do the discovery what setting avoids heartbeat/exit zero status, and really forget for longer :D

--//--

edit: This would be over and above the already available [and by me used], non-BOINC CPU % load control. Mine is set at 40% on Linux and then pauses the client whenever the load is greater... better to loose a few minutes, than loose whole tasks.