Task exited with zero status but no 'finished' file
8 Feb 2011 9:26:25 UTC
Topic 195636
(moderation:
I have had repeated problems recently with Tasks failing to progress and generating the above error message repeatedly, particularly with Tasks Binary Pulsar Search 1.06 (BRP3cuda32fullCPU). The Error Message suggests resetting the Project, which I have tried twice with no improvement.
I am running BOINC Manager 6.10.58 under Ubuntu Linux 10.10 with an Intel Q9950 Quad CPU.
Task exited with zero status but no 'finished' file
)
While it says to reset the project, this usually doesn't fix the problem.
When BOINC finishes with the science program in a normal fashion, it'll exit with a "status 0" and write the contents of what's in memory back to disk (the finished file). Any other number than zero is an error message.
What is causing this exiting with "zero status but no finished file" is usually something else blocking the science program (binary). Think about firewalls, anti-virus or other anti-malware software (though usually those are on Windows), permission problems and such.
How often do those tasks
)
How often do those tasks exit? Those exits can be triggered by changes to the system clock (time server?).
Do you throttle CPU use on that machine (did you set less than "Use at most 100% of CPU time")?
Gruß,
Gundolf
Computer sind nicht alles im Leben. (Kleiner Scherz)
I do have a Firewall (ufw),
)
I do have a Firewall (ufw), but this does NOT affect other Einstein@Home Tasks, or other Projects that I run (Milkyway, Cosmology, SETI etc).
My CPU throttling is set to 50% when active, 100% when inactive; problem certainly occurs when I am not using my PC (inactive).
RE: My CPU throttling is
)
One can only set the amount of memory to use when the PC is active and not active. It cannot be done yet for the CPU, this is only a proposal in the preferences TODO list.
So then you have set other things. Probably the amount of CPU cores your BOINC can use (On multiprocessors, use at most Z% of the processors) and the CPU time (Use at most X% of CPU time). The first one tells BOINC how many cores it can use, the second one at how much it needs to be throttled.
Try to set "Use at most X% of CPU time" to 100. If that fixes it, it is the CPU throttle.
Sorry about my reference to
)
Sorry about my reference to CPU usage; you are (of course) correct that these are Memory usage settings.
"Use at most X% of CPU time" is already set to 100.
I don't know what has happened, as I was definitely getting the Error messages AFTER I reset the Project, but at this moment everything is working correctly - maybe I didn't wait long enough?
I suggest we close this Thread - many thanks for the support.
FWIW -- I experience this
)
FWIW -- I experience this whenever many [?] tasks are being started at the 'same'
time (typically when starting the client). First couple of them almost
instantaneously fail with 'exit code 0 but no finished file' but they get
restarted shortly thereafter and issue doesn't occur any more (unless ofc, client
is stopped and started again).
BOINC 6.10.58, self-built. Fedora 14 x86_64. Not sure if it's related to OP.
My problem ONLY occurs with
)
My problem ONLY occurs with Einstein Binary Radio Pulsar Search 1.06 (BRP3cuda32fullCPU) Tasks; ALL other Tasks, including Einstein Radio Pulsar Search 1.05 (BRP3SSE) function correctly. The Einstein Binary Radio Pulsar Search 1.06 (BRP3cuda32fullCPU) Tasks seem to be the ONLY Tasks which BOINC Manager runs on my PC which use BOTH the CPUs AND my GPU, so maybe the problem is with the GPU specific software and/or the Task.
However, an Einstein Binary Radio Pulsar Search 1.06 (BRP3cuda32fullCPU) Task runs OK if it starts when BOINC Manager starts on starting up my PC in the morning, but when it completes, if I have another Einstein Binary Radio Pulsar Search 1.06 (BRP3cuda32fullCPU) Task waiting to run, this immediately fails. All I do is manually Suspend the Task, and then Resume it just before I shutdown the PC in the evening. This gets round the problem, and allows me to run these Tasks, but is a little irritating!
Hi, I'm seeing the same
)
Hi,
I'm seeing the same problem with BRP 1.08, I run linux with dual gtx 570 graphic cards. Once my tasks are downloaded they instantly go into error:
25-Feb-2011 13:42:34 [Einstein@Home] [error] File PM0049_02211_77.binary has wrong size: expected 4195472, got 4195907
25-Feb-2011 13:42:34 [Einstein@Home] [error] Checksum or signature error for PM0049_02211_77.binary
25-Feb-2011 13:42:34 [Einstein@Home] Computation for task PM0034_030C1.dm_52_2 finished
25-Feb-2011 13:42:34 [Einstein@Home] Output file PM0034_030C1.dm_52_2_0 for task PM0034_030C1.dm_52_2 absent
25-Feb-2011 13:42:34 [Einstein@Home] Output file PM0034_030C1.dm_52_2_1 for task PM0034_030C1.dm_52_2 absent
25-Feb-2011 13:42:34 [Einstein@Home] Output file PM0034_030C1.dm_52_2_2 for task PM0034_030C1.dm_52_2 absent
25-Feb-2011 13:42:34 [Einstein@Home] Output file PM0034_030C1.dm_52_2_3 for task PM0034_030C1.dm_52_2 absent
Here is a link to one of those failed wu:
http://einsteinathome.org/workunit/93696996
Only happens on my linux hosts, the same setup with windows seems to work fine. Any ideas?
Forgot to say, this only
)
Forgot to say, this only happens on linux systems with multiple discrete GPU cards. I tried 2x gtx 8800 and 2x gtx 570 and I had the same problems. However, trying one gtx 295 (two gpus on one board) it worked fine. Removing one of the gtx 570s or 8800s works fine. So it is only in dual configuration with discrete cards. On windows all of the different configurations works. Any ideas? Sorry for spam.
RE: Forgot to say,
)
One thing you can do is look for the EDIT tag at the top of a post as soon as you send it. It will let you edit that post for up to 60 minutes after you originally make the post.