As with the single-machine case, you must run pvfs2-server with a special argument to create the storage space on all the nodes if it does not already exist. Run the following command on every metadata or IO node in the cluster:
root@cluster1# /usr/sbin/pvfs2-server /etc/pvfs2-fs.conf -f
Then once the storage space is created, start the server for real with a command like this on every metadata or IO node in the cluster:
root@cluster1# /usr/sbin/pvfs2-server /etc/pvfs2-fs.conf
If you want to run the server in the foreground (e.g. for debugging), use the -d option.
If you wish to automate server startup and shutdown with rc scripts, refer
to the corresponding section from the single server example.