Next: Stopping and Restarting the
Up: Initializing and Managing a
Previous: The bebopd restart file
  Contents
  Index
- As user root, start a bebopd process on the service node named
in the cplant-host file. Unless you are debugging, start it
with the -d (daemon) argument so it will run in the background.
The bebopd is started at boot time by the user-env
script on the Sandia Albuquerque Cplants.
-
On every compute node in the virtual machine, start a PCT as user
root. Again, unless you are debugging, start it
with the -d (daemon) argument so it will run in the background.
The PCT
looks up the location of the bebopd in the cplant-host file.
It sends an update message to the bebopd at regular intervals until it
receives an acknowledgement. If the PCT does not receive this
acknowledgement from the bebopd within
10 minutes, it gives up and exits.
On the Sandia Albuquerque Cplant, the PCTs are started by the user-env
script at virtual machine boot time.
If you now run pingd on any service node in the virtual machine, you will
see a list of the PCTs that the bebopd has received update messages
from. As far as the bebopd is concerned, this is
the extent of the virtual machine.
Users may now run parallel applications on the listed nodes.
Next: Stopping and Restarting the
Up: Initializing and Managing a
Previous: The bebopd restart file
  Contents
  Index
Lee Ann Fisk
2001-06-25