pssh — run a command on a remote node.
pssh  [ -e envlist] [ -l name] [ -r ] [ -v ] { -n node-id} { -h hostname} [command [args]]
pssh [ -V ]
pssh [ -? ]
pssh runs a command on a remote node. It is similar to the ssh command, except that the ParaStation MPI daemon is used to start the remote command.
-e,
	  --exports=
          envlist
        
            Export the environment variables
            envlist to  the remote command.
            Envlist may be a single
            variable name or a colon separated list of names.
	  
-h,
	  --host=
          node
        
            Run command on node node.
	  
-l,
	  --login=
          name
        
            Login as user name.
            Only root may login as a different user on the node.
	  
-n,
	  --node=
          node-id
        
            Run command on node with ID
            node-id.
	  
-r,
	  --rusage
        Report resource usage upon exit.
-v
        Be more verbose.
-V
        Print version and exit.
-?,
	  --help
        Show a help message.
--usage
        Show a usage message.
      The pssh command runs a shell or a command
      command providing arguments
      args as an admin-task on a remote
      node.
      The local user must be within the adminuser
      list of the ParaStation MPI daemon or must be a member of the
      admingroup list. 
      Refer to 
      psiadmin(8) and ParaStation MPI User's Guide
      for details how to configure
      adminusers or
      admingroups.
    
      Only root may use the -l option to login as a
      different user.
    
      The destination host may be either specified using the hosts
      ParaStation MPI node ID (option -n) or using the
      hosts name (option -h).
    
The admin-task run by pssh will not be counted for the ParaStation MPI process placement and will not be started obeying the placement rules enforced by ParaStation MPI.
To run a serial task using the ParaStation MPI process placement rules, use psmstart(8). To run a command in parallel on multiple nodes, use psh(8). To run a parallel job, use mpiexec(8).