Message boards :
Number crunching :
Wrong exit code?
Message board moderation
| Author | Message |
|---|---|
 Send message Joined: 26 Jul 05 Posts: 63 Credit: 4,083,755 RAC: 0  
|
If a user runs a sub-project job that runs the VM correctly, but CONDOR fails to find and run a single job, that is recorded by BOINC as an error. This is wrong as far as the user and BOINC is concerned - the user has run the VM correctly and it was not his fault that no work for CERN was completed. If these jobs were recorded as a success rather than an error they would show on the statistics pages as the true current status of the sub-project and BOINC users could see which sub-projects to support. |
 Send message Joined: 29 Aug 05 Posts: 1006 Credit: 6,272,128 RAC: 397 |
If a user runs a sub-project job that runs the VM correctly, but CONDOR fails to find and run a single job, that is recorded by BOINC as an error. I'm trying to get this resolved -- I like it as little as you. Some concrete suggestions on how to deal with it would be appreciated, especially from people familiar with the client-BOINC interface. Ideally I'd like to see a backoff of some sort (and hopefully not an "Error while computing" code). Any ideas?  |
|
Send message Joined: 14 Jan 10 Posts: 1280 Credit: 8,489,126 RAC: 1,872  |
I'm trying to get this resolved -- I like it as little as you. Some concrete suggestions on how to deal with it would be appreciated, especially from people familiar with the client-BOINC interface. Ideally I'd like to see a backoff of some sort (and hopefully not an "Error while computing" code). Any ideas? It can be even worse and I've seen that in the past sometimes with my tasks and tasks of other users, that even when several jobs were already successfully done by the VM, the task ended into an error, just because there were no jobs available anymore or there could not made a connection at that moment. That was a long sentence. Now a shorter one: The solution is quite simple. In those cases just write the shutdown file into the shared folder with a valid exit code and BOINC will handle the task successfully. |
©2024 CERN