Phil is correct - this error is usually caused by the executable not
being available on the remote nodes. If you don't have a common file
system, you may want to investigate the -s option to mpirun. It isn't
the most scalable option in the world, but for small clusters it can
work wonders.
Hope this helps,
brian
On Tuesday, August 12, 2003, at 08:33 AM, Phil Ehrens wrote:
> Is that "new program" on an NFS mounted filesystem? If it is,
> it might not show up on all your nodes in a timely fashion.
>
> Did you log into your nodes and try to execute the program by
> hand?
>
> Jose Arreola wrote:
>>
>> I have run some successful tests in my small cluster but with new
>> programs
>> I get the message:
>>
>> mpirun: cannot start "my_program" on n1 (or n2, etc.) : No
>> such file
>> or directory
>>
>> Any suggestions?
> _______________________________________________
> This list is archived at http://www.lam-mpi.org/MailArchives/lam/
>
--
Brian Barrett
LAM/MPI developer and all around nice guy
Have a LAM/MPI day: http://www.lam-mpi.org/
|
