BOINC problem
4 Apr 2006 8:46:31 UTC
Topic 191029
(moderation:
My BOINC manager is acting up weird. When I start it up, it generates the message (previously unknown to me):
"Using earliest-deadline-first scheduling because computer is overcommitted."
It then proceeds to start processing a job and calculating, eating 100% of my CPU, EVEN THOUGH I'M ACTIVELY USING THE PC!
This behaviour started yesterday when my PC crashed in a rather scary way (shutting down completely in an instant, the PC speaker generating high/low beeps). I suspect BOINC is behind this too, because when I rebooted it shut down again (after a few minutes!) and this stopped happening again only after I quit BOINC. I did not check if it is CPU-temperature related, but I prefer not to experiment with these things.
In the logs I find various error messages like:
"2006-04-03 22:46:19 [Einstein@Home] Result r1_0250.0__1526_S4R2a_2 exited with zero status but no 'finished' file"
"2006-04-03 22:56:37 [Einstein@Home] Unrecoverable error for result r1_0250.0__1526_S4R2a_2 (Incorrect function. (0x1) - exit code 1 (0x1))"
I tried a simple re-install (without deleting all files manually), but this doesn't change anything. I guess I can try to delete everything and then reinstall.
My specs:
[*] On which unit did it happen? (See in your Your Account, Results.)
6183915
[*] Which operating system are you using?
WinXP SP2
[*] What make & model of CPU do you have? Is it overclocked?
AMD Athlon XP 2500+, not overclocked
[*] How much memory does your PC have?
256 MB
[*] Did you get the crash when you had graphics/screensaver on?
No
[*] What make & model videocard do you use?
NVidia Magic 128 MB
[*] Is it updated to use the latest DirectX and drivers?
I believe so
[*] Do you have Microsoft .NET installed on your computer?
Don't think so
[*] Are you using the latest BIOS/drivers for your motherboard?
I guess
[*] Which BOINC version are you using? (if need be: And why?)
5.2.13
[*] How is BOINC installed? As a single/shared user install or as a service?
single user
[*] Are you running multiple projects? If yes, list them and tell of the switch between projects time and if you leave the applications in memory when preempted.
Just Einstein@home
[*] When did the crash occur? Did you reboot, was it when changing projects, something else?
1st time: don't know, PC was on, but I wasn't using it
2nd time: a few minutes (<=5) after rebooting
[*] What do you think crashed the Albert unit?
???
Up to yesterday I had set my preferences to have BOINC contact the server every 10 days (it still makes contact every day for some reason, but it has downloaded a lot of jobs), I changed it to 3 days.
I hope someone has any idea. Otherwise I'll just try deleting everything thoroughly, and reinstall.
Stefan.
BOINC problem
)
From the Wiki: Using earliest-deadline-first scheduling because computer is overcommitted.
Ok, this work unit was the
)
Ok, this work unit was the one that crashed due to 'Incorrect function. (0x1) - exit code 1 (0x1)'.
Searching the Boinc Wiki yields this page of explanation. I would go with the first suggestion there and look at this thread to check/re-check install/re-install the video card drivers.
Cheers, Mike.
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
RE: My BOINC manager is
)
This is "normal".
Did you modify your preferences? Check "Your Account" -> "View or edit general preferences" -> "Do work while computer is in use?"
You might want to check your computer stability. Try prime95 and memtest.
This is strange indeed. It usually happens if the application uses instructions not known by the processor, i.e. SSE3 on Pentium I.
Try the above mentioned tools and report back.
Michael
Team Linux Users Everywhere
OK, thanks for all yor help
)
OK, thanks for all yor help and suggestions. I didn't realize that information could be found in the BoincWiki. First place to look next time.
One mystery less: my prefs indeed indicated BOINC to run if the user is active. I don't know how and when this had crept in; I certainly never knowingly set this option.
The "overcommitted" issue probably results from too many jobs being in the queue. I hadn't used that computer for a week, and started to work with it again, maybe that has something to do with it (BOINC suddenly realizing that it has less time to finish all those jobs?). Now I have cancelled all the jobs, and changed the default workload back to 3 days (I had recently increased it to 10). That should take care of it.
Still, I do not understand why it crashed my computer twice. The simultaneous appearance of the "Incorrect function. (0x1) - exit code 1 (0x1)" message and a crash suggest to me that it was a BOINC problem. Now I have updated the video drivers. But note that the PC had been running stable for months (being on all the time) with this version of BOINC and the old video drivers. The screen saver is always off and I didn't use the "Show Graphics" button, so I suspect it is not an OpenGL issue, but that it has something to do with being "overcommitted".
Anyway, now it seems to be back to running smoothly again.
Thanks,
Stefan.
RE: Still, I do not
)
That is likely quite true.
Bear in mind that the BOINC Wiki was derived from accumulated experience of usage reports ( originally compiled by Paul Buck ), and as such does not include all errors or all known errors or lists all causes. It is nonetheless a terrific guide and a longs ways better than none.
Cheers, Mike.
(edit) Thanks and three cheers for the team who moved the Wiki to the new home!!
Top job, looks great!!
I have made this letter longer than usual because I lack the time to make it shorter ...
... and my other CPU is a Ryzen 5950X :-) Blaise Pascal
RE: One mystery less: my
)
If you added a new project recently, it's default may have been run always, and it propagated to all projects (as that is one of the options that does).
Well, you can ignore my
)
Well, you can ignore my previous post. Running the Prime95 Torture Test reveals that my computer is not stable anymore at high workloads. Memory is fine, but CPU temperature appears to be troublesome. This happened after running BOINC continuously and stable for two months.
So BOINC was not to blame, overheating was.
Thanks again for your suggestions,
Stefan.