climateprediction.net (CPDN) home page
Thread 'Repeated \"Unrecoverable errors\"'

Thread 'Repeated \"Unrecoverable errors\"'

Questions and Answers : Unix/Linux : Repeated \"Unrecoverable errors\"
Message board moderation

To post messages, you must log in.

AuthorMessage
old_user163611

Send message
Joined: 16 Feb 06
Posts: 1
Credit: 9,269
RAC: 0
Message 22173 - Posted: 18 Apr 2006, 20:52:20 UTC

After about 2 months trying to contribute to the project, I am giving up. Every work unit has failed with an \"unrecoverable error\", and the software has started a different work unit that has failed a few days or weeks later in the same way. None of the work units have more than 20% completed. As the mean time between failures is a small fraction of the work unit duration, I estimate that I have about a one in a thousand chance of completing any given work unit. At that rate, it is likely to take a few hundred years to complete one work unit.

Unless and until these reliability problems are sorted out, there really is no point in my continuing.
ID: 22173 · Report as offensive     Reply Quote
Les Bayliss
Volunteer moderator

Send message
Joined: 5 Sep 04
Posts: 7629
Credit: 24,240,330
RAC: 0
Message 22179 - Posted: 18 Apr 2006, 22:02:54 UTC

You\'re not going to get far until you increase your memory to the recommended amount of 512 Megs. Especially for the new Coupled Ocean models.


ID: 22179 · Report as offensive     Reply Quote
old_user88305

Send message
Joined: 17 Jul 05
Posts: 1
Credit: 48,798
RAC: 0
Message 22450 - Posted: 28 Apr 2006, 10:01:10 UTC

Hello,

maybe, this link/thread can be a solution:

http://climateapps2.oucs.ox.ac.uk/cpdnboinc/forum_thread.php?id=3846&#20862

Good Luck

Ronsapfel


ID: 22450 · Report as offensive     Reply Quote
old_user10413

Send message
Joined: 3 Sep 04
Posts: 1
Credit: 114,415
RAC: 0
Message 22452 - Posted: 28 Apr 2006, 11:36:07 UTC

Hi, I had a similar problem, and I updated my boinc version to the latest one and it fixed it. Maybe this is related to your problem.
-Tom.
ID: 22452 · Report as offensive     Reply Quote
old_user85713

Send message
Joined: 2 Jul 05
Posts: 8
Credit: 279,985
RAC: 0
Message 22943 - Posted: 27 May 2006, 14:10:05 UTC

Hello,

I have used up several work units the same way, in my case it has something to do with the BOINC \"show graphics\" feature. Somtimes pressing this button leads to a unresolvable computational error, with new BOINC version as well as with the older one.

I then chose to help myself with a simple backup strategy. I usually now start my
laptop X session with a run.sh script located in some backup directory

tar -cjf boinc.tar.bz2 /data/BOINC/
/data/BOINC/run_client > /dev/null &
startx

that will first backup the whole BOINC directory to a bzip2 file
(takes about 5 min) when CPDN is still *not* running, then start boinc
in the background and start the X desktop.

If BOINC chooses to crash unrecoverable because nosy mhofer chose to press
this \"show graphics\" button again AND the machine is luckily offline so it had
no chance to complain at the cliateprediction.net servers (and fetch new work!),
I simply erase the BOINC directory and move the backup over it (expanded with
tar -xvf of course). I start then from the point in time where the backup been
done.

I know this is very cumbersome, but better to lose a work unit due to computation errors that are NOT due to parameter setup of the experiment, but
apparently due to instabilities in the software.

If a model crashes because it is bound to from the experimental setup, the backup would not \"help\" of course. The people responsible for the linux release
whould seriously think about increasing the stability here, and why the graphics
button can lead to an unrecoverable(!) computational error

marcus





ID: 22943 · Report as offensive     Reply Quote
ProfileastroWX
Volunteer moderator

Send message
Joined: 5 Aug 04
Posts: 1496
Credit: 95,522,203
RAC: 0
Message 22951 - Posted: 27 May 2006, 19:25:50 UTC

Hi, Marcus,

Our experience in the Boards suggests many, if not most, graphics-related problems result from out-of-date drivers. (I\'m as guilty as any about not updating drivers in Linux \'cuz it\'s such a bother.) Might be worth the effort, if only to eliminate the possibility.
"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.
ID: 22951 · Report as offensive     Reply Quote
old_user85713

Send message
Joined: 2 Jul 05
Posts: 8
Credit: 279,985
RAC: 0
Message 22972 - Posted: 30 May 2006, 6:19:19 UTC - in response to Message 22951.  

Hi, Marcus,

Our experience in the Boards suggests many, if not most, graphics-related problems result from out-of-date drivers. (I\'m as guilty as any about not updating drivers in Linux \'cuz it\'s such a bother.) Might be worth the effort, if only to eliminate the possibility.



Hi AstroWX,

I did a complete reinstall of BOINC resently, did not change anything (though, it did upgrade the BOINC manager)

Do You mean components of the X server or the KDE desktop, shared libraies BOINC makes use of? I not very skilled with that drivers concept on linux unfortunately

Cheers
Marcus
ID: 22972 · Report as offensive     Reply Quote
ProfileastroWX
Volunteer moderator

Send message
Joined: 5 Aug 04
Posts: 1496
Credit: 95,522,203
RAC: 0
Message 22978 - Posted: 30 May 2006, 23:38:20 UTC - in response to Message 22972.  

Hi AstroWX,

I did a complete reinstall of BOINC resently, did not change anything (though, it did upgrade the BOINC manager)

Do You mean components of the X server or the KDE desktop, shared libraies BOINC makes use of? I not very skilled with that drivers concept on linux unfortunately

Cheers
Marcus


Hi, Marcus,

Your Linux distributor might have drivers on their website; otherwise, you might try the graphics card vendor\'s site. Unfortunately, some manufacturers pretend Linux doesn\'t exist and make life difficult for us re. drivers.

These drivers are at the MotherBoard/BIOS level, as far as I know. Windows often has a driver to allow driver update from within Windows (otherwise, it involved download, unpack and copy to bootable Floppy, boot to the Floppy and continue from there to install). I\'m not aware of any comparable utility in Linux, or KDE, or X -- or the CL.

Hopefully, someone more skilled in Linux than me will come along and lend a hand. (I\'ve used Linux for years because it crashed less often than Windows [no longer true in my experience] and because it gets less attention from crackers. It\'s also better at memory management and doesn\'t require the hateful Defrags. Otherwise, Windows beats it hands down -- much as it distresses me to make that statement. BUT, I use Linux to do work, not to work on Linux.)

"We have met the enemy and he is us." -- Pogo
Greetings from coastal Washington state, the scenic US Pacific Northwest.
ID: 22978 · Report as offensive     Reply Quote

Questions and Answers : Unix/Linux : Repeated \"Unrecoverable errors\"

©2024 cpdn.org