Assuming that your LAM machine has been installed and properly configured, the operation of the system is quite easy. Before you start up the engine, issue the command recon:
gustav@cisr:~/mpi/tutorial/mpi 259 $ recon -v recon: testing n0 (cisr) recon: testing n1 (mercury) recon: testing n2 (venus) recon: testing n3 (earth) recon: testing n4 (mars) recon: testing n5 (jupiter) recon: testing n6 (saturn) recon: testing n7 (uranus) recon: testing n8 (bacchus) recon: testing n9 (ceres) recon: testing n10 (diana) recon: testing n11 (minerva) recon: testing n12 (vesta) gustav@cisr:~/mpi/tutorial/mpi 260 $This command checks if every node of the LAM engine is correctly configured and available. If recon complains that something is wrong, don't try to start LAM. Consult your system administrator, if you can't fix the problem yourself.
Assuming that everything is fine, start the LAM engine by issuing the command lamboot:
gustav@jupiter:~ 201 $ lamboot -v LAM 5.2 - Ohio Supercomputer Center hboot n0 (cisr)... hboot n1 (mercury)... hboot n2 (venus)... hboot n3 (earth)... hboot n4 (mars)... hboot n5 (jupiter)... hboot n6 (saturn)... hboot n7 (uranus)... hboot n8 (bacchus)... hboot n9 (ceres)... hboot n10 (diana)... hboot n11 (minerva)... hboot n12 (vesta)... topology n0...n1...n2...n3...n4...n5...n6...n7...n8...n9...n10...n11...n12... gustav@jupiter:~ 202 $This will take a long while, but once LAM is up, normal MPI applications will start up without such delay.