<!DOCTYPE Article PUBLIC "-//OASIS//DTD DocBook V3.1//EN">
<Article id="index">
   <ArtHeader>
      <Title>pssh HOWTO</Title>

      <pubdate>0.1, 2003-11-13</pubdate>

      <AuthorGroup>
         <AUTHOR>
            <FirstName>Brent</FirstName>
            <SurName>Chun</SurName>
         </AUTHOR>
      </AuthorGroup>

      <RevHistory>
         <revision>
            <revnumber>0.1</revnumber>
            <date>2003-11-13</date>
            <revremark>Initial version.</revremark>
         </revision>
      </RevHistory>

      <abstract>
         <para>
         </para>
      </abstract>
   </ArtHeader>

  <sect1>
    <title>Installation and Setup</title>

    <para>
     To install the software, become <constant>root</constant> on your machine and do
     the following (on RedHat systems):
    </para>

<screen>
   # rpm -ivh pssh-0.2.3-1.i386.rpm
   Preparing...                ########################################### [100%]
      1:pssh                   ########################################### [100%]</screen>

    <para>
     By default, the software installs itself in <constant>/usr/localbin</constant> and
<constant>/usr/local/lib</constant>.  Thus, you'll next want to
modify your <constant>PATH</constant> if needed:
    </para>

<screen>
# export PATH=$PATH:/usr/local/bin</screen>

  </sect1>

  <sect1>
    <title>Preliminaries</title>

    <para>
All four programs will print their usage and give an example
if no arguments are given.  For example, with <constant>pssh</constant>:
    </para>

<screen>
   # pssh
   Usage: pssh [OPTIONS] -h hosts.txt prog [arg0] ..
   
     -h --hosts   hosts file (each line "host[:port] [user]")
     -l --user    username (OPTIONAL)
     -p --par     max number of parallel threads (OPTIONAL)
     -o --outdir  output directory for stdout files (OPTIONAL)
     -t --timeout timeout in seconds to do ssh to a host (OPTIONAL)
     -v --verbose turn on warning and diagnostic messages (OPTIONAL)
     -O --options SSH options (OPTIONAL)
   
   Example: pssh -h ips.txt -l irb2 -o /tmp/foo uptime</screen>

<para>
And for <constant>pscp</constant>:
</para>

<screen>
   # pscp
   Usage: pscp [OPTIONS] -h hosts.txt local remote
   
     -h --hosts     hosts file (each line "host[:port] [login]")
     -r --recursive recusively copy directories (OPTIONAL)
     -l --user      username (OPTIONAL)
     -p --par       max number of parallel threads (OPTIONAL)
     -t --timeout   timeout in seconds to do scp to a host (OPTIONAL)
     -O --options   SSH options (OPTIONAL)
   
   Example: pscp -h hosts.txt -l irb2 foo.txt /home/irb2/foo.txt</screen>

<para>
Note that before using any of these tools, you <emphasis>will need to start
ssh-agent</emphasis>!   This can be done as follows (substitute <constant>zsh</constant>
with your particular shell).
</para>

<screen>
   # ssh-agent zsh
   # ssh-add
   Enter passphrase for /x/bnc/.ssh/identity: </screen>

  </sect1>

  <sect1>
    <title>Examples</title>

  <sect2>
    <title>pssh</title>

    <para>
The following example runs <constant>hostname</constant> on three machines (IPs
or hostnames) specified in the file <constant>ips.txt</constant> using login 
<constant>irb2</constant> and saves the output in <constant>/tmp/foo</constant>.  
</para>

<screen>
   # cat ips.txt
   128.112.152.122
   18.31.0.190
   128.232.103.201
   
   # pssh -h ips.txt -l irb2 -o /tmp/foo hostname
   Success on 128.112.152.122:22
   Success on 18.31.0.190:22
   Success on 128.232.103.201:22
   
   # ls /tmp/foo
   128.112.152.122  128.232.103.201  18.31.0.190
   
   # cat /tmp/foo/*
   planetlab-1.cs.princeton.edu
   planetlab1.xeno.cl.cam.ac.uk
   planetlab1.lcs.mit.edu</screen>

<para>
By default, <constant>pssh</constant> uses at most 32 ssh processes in parallel to
ssh to the various nodes. (This is somewhat important if you're
controlling hundreds or thousands of machines.) By default, it also
uses a timeout of one minute to ssh to a node and obtain a result.
For ssh commands that take longer than this (e.g., <constant>sleep 61</constant>),
the -t option can be used.  Note that <constant>pssh</constant> and 
<constant>pnuke</constant>
have a default timeout of one minute.  
<constant>pscp</constant> and <constant>prsync</constant> have no default timeout, but
one can be specified using the -t option.
</para>

  </sect2>

  <sect2>
    <title>pscp</title>
<para>
Here's an example of using <constant>pscp</constant> to copy files in parallel to
a set of machines.
</para>

<screen>
   # pscp -h ips.txt -l irb2 /etc/hosts /tmp/hosts
   Success on 128.112.152.122:22
   Success on 18.31.0.190:22
   Success on 128.232.103.201:22</screen>

<para>
Using the -r option will perform a recursive copy for copying
entire directories.
</para>

  </sect2>

  <sect2>
    <title>pnuke</title>

<para>
The <constant>pnuke</constant> command is useful when you want to kill a bunch of
processes on a set of machines.  For example, suppose you've got a
bunch of <constant>java</constant> processes running on three nodes that
you'd like to nuke (let's use the three machines from the <constant>pssh</constant>
example).  Here you would do the following:
</para>

<screen>
   # pnuke -h ips.txt -l irb2 java
   Success on 128.112.152.122:22
   Success on 18.31.0.190:22
   Success on 128.232.103.201:22</screen>

<para>
The result of the above is to send <constant>kill -9</constant> to all processes
owned by <constant>irb2</constant> with the string java in their name (as reported
by <constant>ps -ef</constant>).
</para>

  </sect2>

  </sect1>

  <sect1>
    <title>Environment Variables</title>
<para>
All four programs take similar sets of options.  All of these
options can be set using the following environment variables:
</para>

<screen>
   PSSH_HOSTS
   PSSH_USER
   PSSH_PAR
   PSSH_OUTDIR
   PSSH_VERBOSE
   PSSH_OPTIONS</screen>

<para>
Here are some example settings:
</para>

<screen>
   # export PSSH_HOSTS="/x/bnc/ips.txt"
   # export PSSH_USER="irb2"
   # export PSSH_PAR="32"
   # export PSSH_OUTDIR="/tmp/bar"
   # export PSSH_VERBOSE="0"
   # export PSSH_OPTIONS="UserKnownHostsFile /tmp/known_hosts"</screen>

<para>
Using the above settings, the examples can be executed succinctly as:
</para>

<screen>
   # pssh hostname
   Success on 128.112.152.122:22
   Success on 18.31.0.190:22
   Success on 128.232.103.201:22
   
   # ls /tmp/bar 
   128.112.152.122  128.232.103.201  18.31.0.190
   
   # cat /tmp/bar/*
   planetlab-1.cs.princeton.edu
   planetlab1.xeno.cl.cam.ac.uk
   planetlab1.lcs.mit.edu
   
   # pscp /etc/hosts /tmp/hosts
   Success on 128.112.152.122:22
   Success on 18.31.0.190:22
   Success on 128.232.103.201:22
   
   # pnuke java   
   Success on 128.112.152.122:22
   Success on 18.31.0.190:22
   Success on 128.232.103.201:22</screen>

  </sect1>
  <sect1>
    <title>Feedback</title>
<para>
Send me email if you're having problems, find bugs, or have any random
comments: <ulink url="http://www.theether.org">Brent Chun</ulink>
(bnc at theether.org).  Thus
far, I've primarily been testing this software on <ulink
url="http://www.planet-lab.org">PlanetLab</ulink>.
</para>

  </sect1>

</Article>