"Error while computing" on Raspberry Pi
1 Jun 2015 16:43:41 UTC
Topic 198112
(moderation:
After several days of crunching, WU 219617651 ended with "error while computing." Sixteen subsequent work units ended with the same message after only a few seconds of processing; no further work units have been received from Einstein@Home.
Is this a problem that I can correct? If so, how?
Thanks for your help.
"Error while computing" on Raspberry Pi
)
If you click on the task ID on the website you can see the std_err messages that were sent back to the project. The actual error message is
[11:19:41][9484][ERROR] Application caught signal 6.
In other words, there was probably some sort of memory allocation failure or corruption which caused a SIGABRT (signal 6) to be generated and so the application responded by quitting at that point. The interesting thing seems to be that immediately before the reported assertion failure, it looks like the app was restarted from the last saved checkpoint. Did you shut down BOINC and restart it that way or did the app get restarted some other way - perhaps by a running BOINC noticing that it had exited prematurely (without finishing) and deciding to restart it? Whatever the reason, all the subsequent task failures seem to be for the same reason so a reboot to clear corrupted memory might (at least temporarily) fix that.
I'm sure Heinz Bernd (as the author) will respond when he sees your problem with much better advice than I could give. I though I'd suggest a reboot (to completely clear memory) to see if that provides at least a temporary fix. Despite the task failures, you still have a quota of 15 tasks/day so you should be able to immediately get new work after the reboot.
Cheers,
Gary.
Hi Gary--Thank you for your
)
Hi Gary--Thank you for your analysis of the error messages; the info contained in the Task ID meant nothing to me.
I know I didn't shutdown and restart BOINC while the work unit was being processed. The only thing that I can think of that might be related is that I sometimes access the Raspberry Pi from another computer using ssh and vnc to check on the progress of the work unit underway. I also use the Raspberry Pi for other tasks while BOINC is active. Would these uses adversely affect the completion of an Einstein@home task?
As you suggest, I'll reboot the Pi and see if it gets more tasks and runs them successfully. Many thanks for your help.
To follow up on the previous
)
To follow up on the previous thread: I rebooted, but didn't get any new tasks immediately, even after clicking on the "Update" button in BOINC Manager. However, after checking "My Account" several hours later, I see that eight new tasks were sent today and attempted to run, but all failed seconds after starting. The error message is:
[17:46:20][18312][ERROR] Couldn't allocate 50331656 bytes of memory for FFT.
[17:46:20][18312][ERROR] Demodulation failed (error: 1)!
I have the Raspberry Pi running on a powered external hard drive, if that makes any difference.
What should I try to get the Raspberry Pi to process tasks satisfactorily? It completed at least one previously and I have one task pending, waiting for validation.
Any guidance/suggestions on what to try next would be much appreciated. Thanks.