It looks like you need to check your network setup on parmac2 -- LAM
uses standard network name resolution functions to get the hostnames
returned by lamnodes (i.e., we generally store IP addresses, not names;
when we need to report names, we perform an OS query to resolve the
address to a name).
I don't see anything here that would cause a problem. Indeed, if you
can run lamnodes successfully on all nodes, then mpirun'ing an
application should also be able to succeed.
Your initial post said that you got errors saying that LAM was not
running. This does not seem consistent with the information you
provided below. Can you cut-n-paste exactly what you typed and what
happened? (start with running lamboot)
On Jul 1, 2005, at 11:11 AM, Angel Tsankov wrote:
>> - do you have a batch system?
> I'm running a yellowdog linux.
>
>> - how did you run lamboot?
>
> I boot LAM like this: lamboot boot_schema
> Here's the boot_schema file:
> =============================
> parmac3.bas.bg cpu=2
>
> parmac1.bas.bg cpu=2
>
> parmac2.bas.bg cpu=2
>
> parmac4.bas.bg cpu=2
>
> =============================
>> - did lamboot run successfully?
> Yes, it did. At least it did not print any warnings or error messages.
>
>> - can you run "lamnodes" on the node where you ran lamboot?
> Yes, here the result:
> n0 parmac3:2:origin,this_node
> n1 parmac1:2:
> n2 parmac2:2:
> n3 parmac4:2:
>
>> - can you run "lamnodes" on any other node that was lambooted?
> Yes:
> n0 parmac3.bas.bg:2:origin
> n1 parmac1.bas.bg:2:
> n2 parmac2.bas.bg:2:this_node
> n3 parmac2.bas.bg:2: hmm, this should be
> parmac4!!!
>
> n0 parmac3.bas.bg:2:origin
> n1 parmac1:2:this_node
> n2 parmac2.bas.bg:2:
> n3 parmac4.bas.bg:2:
>
> n0 parmac3.bas.bg:2:origin
> n1 parmac1:2:
> n2 parmac2.bas.bg:2:
> n3 parmac4.bas.bg:2:this_node
>
> If you need further info, just ask for it...
> _______________________________________________
> This list is archived at http://www.lam-mpi.org/MailArchives/lam/
>
--
{+} Jeff Squyres
{+} jsquyres_at_[hidden]
{+} http://www.lam-mpi.org/
|
