[Date Prev][Date Next] [Thread Prev][Thread Next] [Date Index] [Thread Index]

PETSc and lam


I just finished making PETSc build with lam (and use the alternatives system
to switch between installed mpich and lam versions :-), and started running
the test suite.  It's having some trouble: it's stuck in the first test with
the CPU running at 90% system, 10% user.  The (Debian-modified) PETSc mpirun
wrapper script runs with options:

     $lam_path/mpirun_lam -w -c $np -s n0 $progname -- $options

where $lam_path is /usr/bin and the other three args are inherited from the
mpirun call.  (I think I attached the whole script a month or so ago.)

strace shows it infinite looping on:

socket(PF_UNIX, SOCK_STREAM, 0)         = 3
connect(3, {sin_family=AF_UNIX, path="/tmp/lam-hazelsct@vaio-s"}, 26) = -1
ECONNREFUSED (Connection refused)
close(3)                                = 0

which is funny, since I can rsh to myself just fine.

Using mpich, it goes through the test suites just fine, but running with
certain options I can get it to stall similarly (and am planning a bug

Eray, which build/runtime options do you use to make PETSc work with lam?  The
ones I include in bmake/linux/base.site?  Do you use the PETSc mpirun.lam, or
just mpirun?


-Adam P.

GPG fingerprint: D54D 1AEE B11C CE9B A02B  C5DD 526F 01E8 564E E4B6

             Welcome to the best software in the world today cafe!

Reply to: