Compute Error when Unit Finished
18 Apr 2010 17:48:38 UTC
Topic 194881
(moderation:
h1_0301.45_S5R4__50_S5GCEa
This job went all the way to the finish but I was registered with a Compute Error. I have never had that happen before. Was it correct?
The points requested matched the 2 who were credited with the unit.
Surely it wouldn't have got to the end with that identical points amount if the computation had gone wrong?
Mike
Compute Error when Unit Finished
)
On all your tasks you can check what messages were sent back to the server. Also on the task with the computation error.
When you scroll all the way down you'll find:
h1_0301.45_S5R4__50_S5GCEa_1_0
-161
Error code -161 means:
"This happens when you have an inconsistent client_state.xml file. Files aren't written to it. Task not found would be the error message."
So that's what happened at the time, the file you uploaded & reported was not present in your client_state.xml file. BOINC uses this file to keep track of just about everything on projects and tasks. Sometimes it can happen, due to unforeseen circumstances, that things aren't written into this file. Or that it went partially corrupt.
The BOINC client only checks the start and end of the file to see if it's still in one piece, not everything in between, as it won't know at any point what should be in there or not. If everything checks out, the file is sane according to BOINC, even if that means that due to a glitch the data about your task wasn't written into this file.
What kinds of glitches? Hard drive problems are the most common. So do a check of your hard drive (with the Windows chkdsk command).
Jord Thanks for replying.
)
Jord
Thanks for replying. I have done Chkdsk without finding anything. Also subsequent jobs have reported correctly.
Mike
Then write it off as one of
)
Then write it off as one of those unexplainable gremlins that happen sometimes with computers. But if it happens again, report back and we can try some other things.
Mine's died twice now upon
)
Mine's died twice now upon completion of a WU, this way:
Mon Apr 19 21:53:28 2010|Einstein@Home|Unrecoverable error for result h1_0453.20_S5R4__60_S5GCEa_1 (process exited with code 4 (0x4))
Closing and re-starting BOINC starts a new job, but it appears that the old one was lost?
RE: Mine's died twice now
)
According to my reading of the FAQ on that error, it could be symptomatic of a problem with the new PPC build. I see the tasks ran for quite a while before expiring—my condolences. :( None of my G4s has got through one so far; they crunch for several projects, so will take a couple more days yet … I hope this isn’t something that always happens right at the end!
Not lost: they were reported as “Client error / Compute errorâ€. Was it necessary for you to relaunch BOINC to start a new task, or was that just precautionary? You might consider setting the project to “No new tasksâ€, just in case—I certainly will.
RE: Mine's died twice now
)
I commented on this in the other thread you posted that problem.
This is definitely not a Mac OS PPC problem.
BM
BM