PXE Boot and John the ripper

Just wanted to relay my experiences with the new PXE boot - distributed john piece. The PXE piece seems to work fine, I can get several other boxes to network boot from the backtrack box. John and mpd on the other hand didn't work so well, I put together the following script to setup the backtrack pxe server so mpd would run correctly. It creates a link for python2.4, creates the mpd.conf file with user selected password, sets the correct permissions for that file and then starts mpd in daemon mode.

Code:

---

#!/bin/bash

if  [ ! -e /usr/bin/python2.4 ]; then
        /bin/ln -s /usr/bin/python2.5 /usr/bin/python2.4
fi

if [ ! -e /etc/mpd.conf ]; then
        echo -n "Enter a password for mpd and press [ENTER] : "
        read password
        /bin/touch /etc/mpd.conf
        echo MPD_SECRETWORD=$password > /etc/mpd.conf
        /bin/chmod 600 /etc/mpd.conf
fi

echo "Starting mpd in daemon mode....."
mpd --daemon --listenport=10000

---

In order for the PXE booted workstations to connect properly to the mpd server I had to make a couple changes. These include setting the correct 192.168.1.x ip address in the /etc/hosts file, creating the /etc/mpd.conf file with the right MPD_SECRETWORD password and then starting mpd with mpd --daemon --ncpus=<number of cpus on the box> -h 192.168.1.1 -p 10000

That gets the mpd piece working, however, i could not get john working with the currently installed john versions. It seemed that neither the pxe booted workstations or the pxe boot server had the right mpich patched version of john. Once I got the patched version from hxxp://bindshell.net/tools/johntheripper/john-1.7.2-bp17-mpi6.tar.gz and put it on each of the workstations that I wanted to be included in the cluster things started working. I just needed to make sure the hash that I wanted to crack was on each box and that the path to john was the same across the board.

Hope that helps anyone that has been struggling with this.

-Trip

Your script took care of the python issue (thanks). BUT on the client nodes, while booting off the network, the process crashes with the following error:-

*creating /etc/fstab
linuxrc: linuxrc: 188: cannot create union/etc/fstab: Directory nonexistant
GFS2: path_lookup on /mnt/boot returned error
GFS2: gfs2 mount does not exist
*changing root directory...
Fatal error occured - Can't find executable chroot command
*Something went wrong and we can't continue. This should never happen.
*Please reboot your computer with Ctrl+Alt+Delete ...
/union#

Help...

Tripleboc,

Thx for you hek-lp, i have tried dnetj, but it is a little bit solw, so I will try mpi.

So if I have understood :

I have to compile the mpi tar.gz on each client, and use the backtrap as a mpi server.

Don u think taht it is compulsory for me to use PXE, could I use only dhcpd on my lan.

Asdp38