Casiano Rodriguez-Leon


parpush - Secure transfer of files between clusters via SSH


  parpush [options] '*.tar.gz m1:*.pl m2:*.txt' c1%c2:/dir1 c3+c4:/dir2 ... 

  # You need to set first a configuration file describing the
  # set of clusters: 

      $ cat $HOME/.clustersrc
      cluster1 = machine1 machine2 machine3
      cluster2 = machine2 machine4 machine5
      num =
      # Comments are allowed
      # same as range = cc100 cc101 cc102
      range = cc100..102

  # or have the cluster defined in the $HOME/.csshrc file associated with cssh.


 --configfile file 
                    File describing the clusters. If not specified
                    parpush will look for the files:

 --scpoptions 'options for scp'
                     A string with the options for scp.
                     The default is no options and '-r' if 
                     sourcefile is adirectory

 --program '/usr/local/bin/scp'
                     A string with the name of the program to 
                     use for secure copy. The default is 'scp'

 --name  machine1=string1 --name machine2=string2
                    It may appear several times.
                    The value associated used with machine1
                    will be used  in the @# and @= macros 
                    instead the machine name. 

                    Maximum number of concurrent processes


                    runs cssh to the target machines

                    It shows the scp commands that will be issued
                    but does not transfer the files



  # Copy 'sourcefile' to the union of cluster1 and cluster2
  $ parpush sourcefile  cluster1+cluster2:/tmp 

  # Copy 'sourcefile' to the /tmp/ directory in machine1, machine2, machine3
  # machine4 and machine 5
  $ parpush sourcefile machine1..5:/tmp/  

  # Copy 'sourcefile' to the intersection of cluster1 and cluster2
  # i.e. to 'machine2'
  $ parpush sourcefile  cluster1*cluster2:/tmp 

  # Copy 'sourcefile' to the machines in cluster1 that don't belong to cluster2
  # i. e. to 'machine1', 'machine2'
  $ parpush sourcefile  cluster1-cluster2:/tmp 

  # Copies file 'sourcefile' to file 'tfmachine1.txt' in 'machine1'
  # and to 'tfmachine2.txt' in 'machine2'. The macro '@=' inside
  # a path is substituted by the name of the target machine
  $ parpush sourcefile  cluster1-cluster2:/tmp/tf@=.txt

  # Copies all the files with name '.bashrc' in the home directories
  # of machines in cluster1 to the '/tmp/' directories of machines
  # in cluster2. The macro '@#' stands for the name of the source machine.
  # Thus, the file .bashrc in machine1 will be copied as 
  # '/tmp/bashrc_at_machine1' in the machines in 'cluster2':
  $ parpush -v cluster1:.bashrc cluster2:/tmp/bashrc_at_@#

  # Copies .bashrc in machine1 and .bashrc in machine2
  # to machine3:/tmp/bashrc.ORION and machine3:/tmp/bashrc.BEO
  $ parpush -n machine1=ORION -n machine2=BEO \
            'machine1:.bashrc beowulf:.bashrc' \

  # A more complicated formula.
  # Though 'machine2' is an alias for, (see the
  # cluster definition above) they aren't 
  # considered equal by parpush. The file will be transferred to 
  # 'machine2'
  $ parpush 'sourcefile'  '(cluster1+cluster2)-num:/tmp'

  # Several cluster expressions may appear as targets.
  # Send 'sourcefile' to machines in cluster1 but not in cluster2 
  # and store it in /tmp. Send it to machines in cluster2 but not cluster1
  # and store it at the home directory
  $ parpush sourcefile  cluster1-cluster2:/tmp  cluster2-cluster1:

  # Copy from remote machine 'machine1' the file 'file.txt' to
  # all the machines in 'cluster1' other than 'machine1':
  $ parpush machine1:file.txt cluster1-machine1:

  # You can also transfer several files from several source clusters/machines to
  # some set of machines. In such case
  # protect the source with single quotes. 
  $ parpush  'machine1:file1.txt machine2:file2.txt' cluster1-machine1-machine2:

  # You can "rename" the names of the source machines. In this example
  # the file '.bashrc' at machine1 will be copied as 'dog_bashrc' in
  # all the machines in 'cluster2' (other than 'machine1'):
  $ parpush -n machine1=dog -n machine2=cat -n machine3=mouse -v cluster1:.bashrc \

  # A combination of local and remote files can be sent. 
  # Protect the source with single quotes. The following command
  # sends 'localfile.txt' in the local machine and 'remote.txt' in 'machine4'
  # to machine3
  $ parpush  'localfile.txt machine4:remote.txt' cluster1-machine1-machine2:

  # Globs can be used in the sourcefile argument.
  # All the files matching the glob 'file*' in the local machine will be sent.
  # Also those in machine1 matching the glob '*.pl'
  $ parpush  'file* machine1:*.pl' cluster2-machine2:/tmp

  # All the files matching the glob 'file*' in the 'machine1' will be sent
  # to the local machine.  The directory 'dir/' in machine2 will be also sent
  # to '/tmp/dir/' in the local machine.
  $ parpush  'machine1:*.pl machine2:dir/' :/tmp

  # The macro '@#' stands for the "source machine". Thus, in the example
  # below file 'file.txt' in machine1 will be copied to file '/tmp/file.txt.machine1'
  # in machine3. File 'file.txt' in machine2 will be copied to 
  # file '/tmp/file.txt.machine2' in machine3.
  $ parpush 'machine1:file.txt machine2:file.txt' machine3/tmp/file_txt.@#  

  # You have to write a colon for any target, even if the target is the local host.
  # in the example below 'file.txt' in machine1 will be copied to 
  # file '/tmp/file.txt.machine1' in the local machine. File 'file.txt' in machine2 
  # will be copied to file '/tmp/file.txt.machine2' in the local machine
  $ parpush 'machine1:file.txt machine2:file.txt' :/tmp/file_txt.@#  


Install Set::Scalar first. Then the installation uses the traditional procedure. The program cssh (clustercssh) is not needed but I recommend its installation. Then issue the usual commands (or use cpan):

   perl Makefile.PL
   make test
   make install


parpush push files and directories across sets of remote machines.

Syntax of Cluster Description Files

Unless the option --configfile is specified parpush will look for a filename named ~/.clusterrc in the home directory. If it does not exists, the looks for ~/.csshrc. Last, it looks for the file /etc/clusters.

      $ cat Cluster
      cluster1 = machine1 machine2 machine3
      cluster2 = machine2 machine4 machine5
      num =
      range = cc10..11a4..5

Ranges are allowed. Thus, the definition above

      range = cc10..11a4..5

is equivalent to:

      range = cc10a4 cc10a5 cc11a4 cc11a5

See man cssh to find out how to describe a cluster in the ~/.csshrc file.

parpush Syntax

When calling parpush you have to specify the source and the targets. Each target is split in two parts: the cluster description and the path. You have to write a colon for any target, even if the target is the local host. This behavior differs from scp. in the example below 'file.txt' in machine1 will be copied to file '/tmp/file.txt.machine1' in the local machine. File 'file.txt' in machine2 will be copied to file '/tmp/file.txt.machine2' in the local machine

  $ parpush 'machine1..2:file.txt' :/tmp/file_txt.@#  

The Syntax of Cluster Expressions

  • s + t union

  • s * t intersection

  • s - t difference

  • s % t symmetric_difference

  • Ranges are allowed. Use .. to define them. Thus cc101..102.4..5 defines the set of machines


Path Syntax. The @= macro

Inside a path the macro @= stands for the name of the current machine. Thus, the command:

  $ parpush file.txt machine1+machine2:/tmp/@=.txt

copies file.txt to machine1.txt in machine1 and to machine2.txt in machine2.

Path Syntax. The @# macro

Inside a path the macro @# stands for the name of the source machine. Thus, the command:

  $ parpush 'machine1:file.txt machine2:file.txt' :/tmp/file_txt.@#  

copies file.txt in machine1 to /tmp/file.txt.machine1 in the local machine and the file with the same name in machine2 to the file /tmp/file.txt.machine2.

Source Syntax

If your source is a file or directory nothing is needed. If you are going to send several files you must protect them inside single quotes as in:

  $ parpush  'machine1:file1.txt machine2:file2.txt' cluster1-machine1-machine2:

cluster expressions can be used in the source description argument as in:

  $ parpush -v 'machine1+machine2:.bashrc' machine3:/tmp/bashrc_at_@#
  Executing system command:
          scp  machine1:.bashrc machine3:/tmp/bashrc_at_machine1
  Executing system command:
          scp  machine2:.bashrc machine3:/tmp/bashrc_at_machine2
  machine3 output:

  machine3 output:


To use this script you have to set automatic authentication via SSH between the source machine (your machine) and the other destiny machines. This section explains the simplified procedure.

SSH includes the ability to authenticate users using public keys. Instead of authenticating the user with a password, the SSH server on the remote machine will verify a challenge signed by the user's private key against its copy of the user's public key. To achieve this automatic ssh-authentication you have to:

  • Generate a public key use the ssh-keygen utility. For example:

      local.machine$ ssh-keygen -t dsa -N ''

    The option -t selects the type of key you want to generate. There are three types of keys: rsa1, rsa and dsa. The -N option is followed by the passphrase. The -N '' setting indicates that no passphrase will be used. This is useful when used with key restrictions or when dealing with cron jobs, batch commands and automatic processing which is the context in which this module was designed. By default, in OpenSSH, your identification will be saved in a file /home/user/.ssh/id_dsa. Your public key will be saved in /home/user/.ssh/

  • If still you don't like to have a private key without passphrase, provide a passphrase and use ssh-agent to avoid the inconvenience of typing the passphrase each time. ssh-agent is a program you run once per login session and load your keys into. Keys are added to the ssh-agent using the program ssh-add. From that moment on, any ssh client will contact ssh-agent and no more passphrase typing will be needed. Use the program keychain to manage the communication with the agent from different sessions.

  • Once you have generated a key pair, you must install the public key on the remote machine. To do it, append the public component of the key in


    to file


    on the remote machine. If the ssh-copy-id script is available, you can do it using:

      local.machine$ ssh-copy-id -i ~/.ssh/ user@remote.machine

    Alternatively you can write the following command:

      $ ssh remote.machine "umask 077; cat >> .ssh/authorized_keys" \
                                      < /home/user/.ssh/

    The umask command is needed since the SSH server will refuse to read a /home/user/.ssh/authorized_keys files which have loose permissions.

  • Edit your local configuration file /home/user/.ssh/config (see man ssh_config in UNIX) and create a new section for GRID::Machine connections to that host. Here follows an example:

     # A new section inside the config file: 
     # it will be used when writing a command like: 
     #                     $ ssh gridyum 
     Host gridyum
     # My username in the remote machine
     user my_login_in_the_remote_machine
     # The actual name of the machine: by default the one provided in the
     # command line
     # The port to use: by default 22
     Port 2048
     # The identitiy pair to use. By default ~/.ssh/id_rsa and ~/.ssh/id_dsa
     IdentityFile /home/user/.ssh/yumid
     # Useful to detect a broken network
     BatchMode yes
     # Useful when the home directory is shared across machines,
     # to avoid warnings about changed host keys when connecting
     # to local host
     NoHostAuthenticationForLocalhost yes
     # Another section ...
     Host another.remote.machine an.alias.for.this.machine
     user mylogin_there

    This way you don't have to specify your login name on the remote machine even if it differs from your login name in the local machine, you don't have to specify the port if it isn't 22, etc. This is the recommended way to work with GRID::Machine. Avoid cluttering the constructor new.

  • Once the public key is installed on the server and the key added to the agent, you should be able to authenticate using your private key

      $ ssh remote.machine
      Linux remote.machine 2.6.15-1-686-smp #2 SMP Mon Mar 6 15:34:50 UTC 2006 i686
      Last login: Sat Jul  7 13:34:00 2007 from local.machine



Casiano Rodriguez-Leon <>


Copyright (C) 2009-2009 by Casiano Rodriguez-Leon

This library is free software; you can redistribute it and/or modify it under the same terms as Perl itself, either Perl version 5.8.8 or, at your option, any later version of Perl 5 you may have available.