On Jun 10, 2005, at 4:33 PM, Paul Mitchell wrote:
> We're attempting to use Platforms LSF with a set of SO X Xserves (and
> it's been an uphill battle, let me state). We're very close to making
> it
> work, save for one confusing aspect of lamboot. Here's the scenario:
>
> 1) WE never managed to get openssh to work in passwordless
> authentication
> mode,
Bummer. There's no reason it shouldn't work on OSX, however; we do it
all the time here (we have a small OSX development cluster).
> so I configured lam as follows:
>
> configure --build=powerpc --with-rsh=/usr/local/bin/lsgrun.sh
> --without-romio
This is probably better, anyway, because you'll get better LSF control
and accounting statistics from your jobs.
>>> [snipped]
>>> LAM tried to use the remote agent command "ssh"
>>> to invoke the following command:
>>>
>>> ssh bp03 -n '( ! [ -e ./.profile] || . ./.profile;' hboot -t
> -c
>>> lam-conf.lamd -s -I '"-H 152.2.3.40 -P 59910 -n 1 -o 0"' )
>
> My understand is that lamboot tries to do run two things (separated by
> ;)
> 1) if .profile exists, then source it
> 2) run hboot.
>
> However I think lamboot sets the command line syntax wrongly as we can
> see the first single qoute "'" after -n has no match at all, so does to
> "(".
>
> SO I've been looking through the source for LAM-7.1.1, and I can find
> the
> string:
>
> ( ! [ -e ./.profile] || . ./.profile
Yes, this is unfortunately a known problem in 7.1.1. It's fixed in the
7.1.2 betas -- can you give that a whirl?
http://www.lam-mpi.org/beta/
> in laminfo and lamboot, but nowhere in any of the .c or .cc files.
> Where
> is this string defined and is there some way that I can correct the
> mis-call?
If you still care, that string is assembled in
share/ssi/boot/rsh/src/ssi_boot_rsh_inetexec.c
--
{+} Jeff Squyres
{+} jsquyres_at_[hidden]
{+} http://www.lam-mpi.org/
|
