FWIW, LAM silently ignores when you provide the hostfile in a PBS job;
it always uses the tm stuff.
I don't know why offhand lamboot is doing that; is there any chance
you can upgrade to Open MPI? It has TM support as well.
On Sep 24, 2008, at 11:01 AM, Sean Reque wrote:
> I tried searching for this in the archives but the search function
> on the web site seems broken.
>
> When I try to use LAM MPI with torque, I get an error message saying
> the following: "The boot SSI tm module found that your local host is
> not in the
> hostfile "PBS_NODEFILE"." This happens when I run lamboot and if I
> directly specify the nodes file with lamboot $PBS_NODEFILE.
>
>
>
> However, when I cat the nodes file I see plainly that node n27 is in
> fact in the file and when I run hostname it verifies that I am on
> host n27. When I run lamboot with -d, I get the following:
>
> n-1<18466> ssi:boot:tm: found the following 1 hosts:
> n-1<18466> ssi:boot:tm: n0 n27 (cpu=1)
>
> What is the tm module looking for that isn't being correctly
> provided? I am running lam 7.1.4 with torque 2.3.4.
>
>
> _______________________________________________
> This list is archived at http://www.lam-mpi.org/MailArchives/lam/
--
Jeff Squyres
Cisco Systems
|
