[mvapich-discuss] MVAPICH2-1.6 issues with /usr/bin/at command?

[Diaz, DavidA (Plymouth)]

Hello:



Ive built mvapich2-1.6rc1 and have it running on my Linux cluster here.



I seem to have a problem using mvapich2-1.6 with the /usr/bin/at command.  My mvapich2-1.6 build works fine outside of using /usr/bin/at.

Do you know of any issues/bugs with version 2-1.6 and the at command?



BTW:I don't have an issue at all with mvapich-1.2 using the at command.



LD_LIBRARY_PATH is properly set to find the libib libraries (including libibumad.so.2) in /usr/local/lib64



Here are some screen results:



jdoe at hostname:/home/jdoe> /usr/bin/at now

warning: commands will be executed using /bin/sh

at>/usr/local/mvapich2-1.6/bin/mpirun_rsh -ssh -np 2 n001 n002 /bin/ls /tmp >& /tmp/dave.out

at><EOT>

job 451 at 2011-05-17 13:04



jdoe at hostname:~>cat /tmp/dave.out

/usr/local/mvapich2-1.6/bin/mpispawn: error while loading shared libraries: libumad.so.2:

cannot open shared object file: No such file or directory

/usr/local/mvapich2-1.6/bin/mpispawn: error while loading shared libraries: libumad.so.2:

cannot open shared object file: No such file or directory

child_handler: Error in init phase...wait for cleanup! (0/2mpispawn connections)

child_handler: Error in init phase...wait for cleanup! (0/2mpispawn connections)

Failed in initialization phase, cleaned up all the mpispawn!



When I test mvapich2-1.6 on the command line it works just fine...

LD_LIBRARY_PATH is properly set to find the libib libraries (including libibumad.so.2) in /usr/local/lib64



jdoe at hostname:/usr/local/mvapich2-1.6/bin/mpirun_rsh -ssh -np 2 n001 n002 /bin/ls /tmp > /tmp/dave.out

/usr/X11R6/bin/xauth: error in locking authority file /home/jdoe/.Xauthority

jdoe at hostname:/home/jdoe>



jdoe at hostname:~> cat /tmp/dave.out

file1 listing from node1

file2 listing from node 1

file1 listing from node 2

file2 listing from node 2

When I try mvapich-1.2 with the at command, it works just fine...



jdoe at hostname:/home/jdoe> /usr/bin/at now

warning: commands will be executed using /bin/sh

at>/usr/local/mvapich-1.2/bin/mpirun_rsh -ssh -np 2 n001 n002 /bin/ls /tmp >& /tmp/dave.out

at><EOT>

job 452 at 2011-05-17 13:11



jdoe at hostname:~> cat /tmp/dave.out

file1 listing from node1

file2 listing from node 1

file1 listing from node 2

file2 listing from node 2



Is there a known bug with mvapich2-1.6 using the /usr/bin/at command in Linux?







Thanks much!



David A. Diaz

Information Systems Security Officer

ATK Advanced Weapons

Plymouth, MN

(763) 744-5003


-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://mail.cse.ohio-state.edu/pipermail/mvapich-discuss/attachments/20110518/1c32456a/attachment.html