On Sat, 17 Apr 2004, Kartik Srinivas wrote:
> I have a two computer cluster setup. I am getting errors when running
> the lamtests-7.0.4, the lamboot command works and I am able the see the
> nodes. I have attached a zip file containing all the files mentioned in
> the README.
Sorry for taking so long to reply; the end of the semester got *quite*
hectic. :-(
The only errors I see in your output have to do with the MPI-2 dynamic
functions: spawn and join.
The spawn problems, as noted by the error message, are typically caused by
a lack of a uniform filesystem. The test first checks to see if the
executable can be found on all nodes in the process. If it can't find it
on all nodes, it prints out that message and then skips the test.
The join test first makes a socket between adjoining ranks in
MPI_COMM_WORLD. It does this by making a "best guess" at what the IP
addresses are of machines in the test. Sometimes, in oddly-configured
networking environments, it will guess wrong (it resolves the hostname to
an IP address), and the connect() will therefore fail.
Neither of these errors are problematic; I'm guessing it's just the
peculiarities in your environment that prevent the tests from running
properly. I wouldn't worry about them.
--
{+} Jeff Squyres
{+} jsquyres_at_[hidden]
{+} http://www.lam-mpi.org/
|
