Hello,
I faced a problem.
Sometimes simulations stopped and some message appeared without error files.
There is no error files and no error message in calculation log.
Do someone know about this?
Best regards,
Tsubasa
Error without error files
Forum rules
Please read the forum usage recommendations before posting.
Please read the forum usage recommendations before posting.
-
- Posts: 4069
- Joined: Mon Feb 20, 2012 3:25 pm
Re: Error without error files
Hello,
Though this might not a defined by a norm, it seems many MPI environments use the signal number that crashed the compilation from a single process + 128 to define the returned signal number.
So 137 would be 128 + 9, meaning the computation is killed by signal 9 (same as "kill -9", see also /usr/include/bits/signum.h on must systems).
This signal cannot be intercepted, so the code does not have the time to write a backtrace.
This probably means you system or batch scheduler is killing the computation. It might have run out of allocated time.
If you have no batch scheduler (probably not on a simple virtual machine), if the memory requirements for the code exceed that allocated to the virtual machine, you might also have this error, or a cleaner stop, depending on how fast memory requirements grow. Not sure what other conditions can cause this.
Best regards,
Yvan
Though this might not a defined by a norm, it seems many MPI environments use the signal number that crashed the compilation from a single process + 128 to define the returned signal number.
So 137 would be 128 + 9, meaning the computation is killed by signal 9 (same as "kill -9", see also /usr/include/bits/signum.h on must systems).
This signal cannot be intercepted, so the code does not have the time to write a backtrace.
This probably means you system or batch scheduler is killing the computation. It might have run out of allocated time.
If you have no batch scheduler (probably not on a simple virtual machine), if the memory requirements for the code exceed that allocated to the virtual machine, you might also have this error, or a cleaner stop, depending on how fast memory requirements grow. Not sure what other conditions can cause this.
Best regards,
Yvan