[MPICH] mpich2 & gdb on solaris issues

Gunter, Tim tgunter at arete.com
Fri Jan 13 14:55:46 CST 2006


I have been having problems running jobs through gdb on a cluster of solaris 9 machines.

When I run my jobs w/o specifying "-gdb", they run to completion as expected.  When I try to use gdb, I never get a gdb prompt until I issue a "z" command.  When I enter a gdb command(either after issuing a "z" command and receiving a gdb prompt, or without a prompt) nothing is printed to the screen.  No output ever appears from any of the gdb commands I issue, and the job hangs, and never terminates.  The following is the output from one of my attempts to use gdb.  The mpi ring I used for this session only had one cpu in it, but the behavior is the same when I use a ring of four machines.

=========================
gunter/experimental/MPI $ mpiexec -n 4 ./cpi     
Process 0 of 4 is on Sun1
Process 1 of 4 is on Sun1
Process 2 of 4 is on Sun1
Process 3 of 4 is on Sun1
pi is approximately 3.1415926544231239, Error is 0.0000000008333307
wall clock time = 0.154744
gunter/experimental/MPI $ mpiexec -gdb -n 4 ./cpi
l


b 30
r
z
0-3:  (gdb) l
z
0-3:  (gdb) b 30
r
z
0-3:  (gdb) r
bt
z
0-3:  (gdb) bt
========================

Any help, or ideas on how to debug this further would be greatly appreciated!

Thanks

--Tim
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.mcs.anl.gov/pipermail/mpich-discuss/attachments/20060113/730f4cda/attachment.htm>


More information about the mpich-discuss mailing list