Questions and Answers :
Windows :
White Screen at 99,99% - Client error
Message board moderation
Author | Message |
---|---|
Send message Joined: 13 Jan 08 Posts: 5 Credit: 18,355 RAC: 0 |
I just finished a WU, but it says Client error. http://climateapps2.oucs.ox.ac.uk/cpdnboinc/results.php?hostid=821907 At 99,99% I switched on the screensaver, because it stopped. Earth was completly white. I restarted BOINC, but then it said \"Berechnungsfehler\". I don\'t no what it says in the English Version, but in means that an error occurred while calculating. Can anyone tell me what happend, and if the results will still be used? |
Send message Joined: 9 Jan 07 Posts: 467 Credit: 14,549,176 RAC: 317 |
The model, 7098535, was very near the end, at which point a significant amount of processing occurs. If the PC has a problem at that point then the model may rewind or crash. The error report says \"Incorrect function. (0x1) - exit code 1 (0x1)\" (auf Deutsch, Unzulässige Funktion. (0x1) - exit code 1 (0x1)). There is a discussion on the BOINC Wiki here. It may be that starting the graphics application while the model was post-processing may have caused the crash. The computer seems to be a laptop (Intel T2300 processor) and has slightly less than 512 MB of RAM, which together suggest an integrated graphics device - the graphics might not have been able to start properly due to a memory shortage. The reporting of trickles is slow at present and it may therefore be possible that more trickles will appear over the next few days. Otherwise, unless you have a backup to restore, the results for the final phase will have been lost. Very bad luck so close to the end. |
Send message Joined: 13 Jan 08 Posts: 5 Credit: 18,355 RAC: 0 |
Before it crashed it did trickle-up. I hope that the Results will be received. This was my first climateprediction.net-WU, so I didn\'t know it does post-processing, I thought it crashed. Other BOINC projects don\'t do that, or it doesn\'t take that long, and you don\'t notice it. Next time I won\'t switch the screensaver on. ()-;) |
Send message Joined: 5 Sep 04 Posts: 7629 Credit: 24,240,330 RAC: 0 |
While a model is running, it saves different types of info in different files. At the end of each phase, (for a slab model), or at the end of the model for the other types, these files are integrated, and finally zipped. Then a last trickle for the phase, plus the zip file, are uploaded. This process can takes 10 or more minutes to complete, and you\'ll see the cursor changing to an hour glass and back to an arrow a lot during this. Interrupting it at this point can be fatal. Also, there is a minor problem with the current public slab models, in that there are over a hundred trickle_up_ files created at end of phase, and it\'s the last one that\'s important. This has been fixed in a new type of 4 phase slab model currently being tested. Backups: Here |
Send message Joined: 13 Jan 08 Posts: 5 Credit: 18,355 RAC: 0 |
I never got an hourglass, i had always the normal arrow. I remember that before I started the screensaver I opened the taskmanager, and the CPDN-process had 20%, and Einstein@home had 51%. The CPDN-process usually had about 50%. If it\'s working so hard that it even shows you the hourglass shouldn\'t it be more than 20%? I think it doesn\'t show you the hourglass if the Idle-process is a little fewer than 30%. |
Send message Joined: 5 Sep 04 Posts: 7629 Credit: 24,240,330 RAC: 0 |
Taking into account that there\'s a wide range in processor speeds, programs running, etc, I\'m be a bit more general, (and vague), and say that the cursor flickers a lot. And, if by \"screensaver\" you mean the globe display that appears when you click the Show graphics button, this is NOT a screensaver. There IS a screensaver included with the package, as some people have insisted on one, but this is dynamic, unlike for example, the Windows ss; (it\'s constantly being updated from the running model), and it\'s reccomended NOT to use it, as it can take up a lot of processor time. Backups: Here |
Send message Joined: 9 Jan 07 Posts: 467 Credit: 14,549,176 RAC: 317 |
One further trickle has arrived! One more to go ... |
Send message Joined: 13 Jan 08 Posts: 5 Credit: 18,355 RAC: 0 |
I guess the last trickle won\'t arrive. Too bad. |
Send message Joined: 13 Jan 06 Posts: 1498 Credit: 15,613,038 RAC: 0 |
Doesn\'t look like it. The models really don\'t like to be interrupted during the final post-processing phase, they are vulnerable then. During this step the globe will not be shown normally since this can only be displayed when the climate model itself is running. I seem to recall Carl or Tolu had fixed the flashing-cursor problem during the previous round of beta tests, which probably explains why the cursor was normal. I'm a volunteer and my views are my own. News and Announcements and FAQ |
©2024 climateprediction.net