Hi,
I have setup a 5 nodes cluster (1 master, 4 slaves) using Oscar 2.2.
After installation, I logon to the head node and recon, lamboot and test
a simple mpi program successfully. After switching off and restart the
machines, at the head node, whenever I recon or lamboot, I was asked for
the password of the slave node. Below are the messages:
lamboot -v lamhosts4
LAM 6.5.9/MPI 2 C++/ROM - Indiana University
Executing hboot on n0 (lilacnode1.lilac.com -1 CPU) ....
Password:
permission denied, please try again (this message appear when I press
enter, and I would have thought that Oscar would have setup passwordless
logon for the cluster, still don't know why)
-----------
I also did a recon in another instance:
recon -v lamhosts4
recon: --testing n0 (lilacnode1.lilac.com)
password:
when I gave the correct password, the following message appear:
"Could not chdir to home directory /home/chiam; no such file or
directory "
Would really appreciate any help to get my lam-mpi working.
Thanks and best regards.
Chiam Tow Jong
Experimental Mechanics Laboratory
Mechanical Engineering Department
National University of Singapore
_____
Upgrade Your Email - Click here!
<http://promos.hotbar.com/promos/promodll.dll?RunPromo&El=hotbar%5feleme
nt%3bst%3b&SG=&RAND=68174>
|