Beowulf cluster using SuperRescue Bootable CD’s as client nodes.

Background:

I was experimenting with H. Peter Anvin’s excellent SuperRescue CD and wondered if I could make one that would allow me to create a Beowulf cluster node that I could boot up only when I needed the cluster. In an office environment I would be able to put the cd’s in coworkers machines when they left for the evening, reboot and run some Cluster applications. This also allows me to have a single cd that I can just copy to create the clones for each node.

This is how I did it the first time. I also wanted to make as few modifications to HPA’s SuperRescue as possible. MUCH THANKS to Bob Tracy for helping me with ALL the Linux stuff. Without whom I’d still be beating my head against every wall around.

This project has a long way to go so I’m open to suggestions and corrections. Email me at steve@brim.net.

 

1 Download and read ALL about version 1.3 of SuperRescue from http://www.kernel.org/pub/dist/superrescue/ pay special attention to what you must do to create your own version of a SuperRescue cdrom.

2 Create a Beowulf Host computer. I used a RedHat 6.2 machine with PVM installed from the RPM.

3: Create new user on the Host computer named "user" with UID of 500 and GID of 500.

4: Set up a DHCP server for use by beowulf nodes.

5 Setup the DNS server with names for each node.

The following steps are made to the SuperRescue CD files loaded on a separate machine under the ~/buildroot directory.

6.Add the PVM rpm’s.

rpm –ivh pvm(verison).rpm

7 Add a .rhosts file to ..buildroot/root/home/user/

8. Add

"export PVM_ROOT=/home/smith/pvm3
source $PVM_ROOT/lib/bashrc.stub"

to ../buildroot/root/home/user/.bashrc reference http://www.informatik.uni-stuttgart.de/ipvr/bv/cppvm/online-doc/node8.html

9. Add ../buildroot/root/etc/pam.d/rlogin file, make sure you do not require secure tty when logging into the nodes. To fix this, switch the order of the modules pam_securetty.so and pam_rhosts_auth.so., and end up with this:

auth

sufficient

/lib/security/pam_rhosts_auth.so

auth

required

/lib/security/pam_securetty.so

auth

required

/lib/security/pam_pwdb.so shadow nullok

auth

required

/lib/security/pam_nologin.so

account

required

/lib/security/pam_pwdb.so

password

required

/lib/security/pam_cracklib.so

password

required

/lib/security/pam_pwdb.so shadow nullok use_authtok

session

required

/lib/security/pam_pwdb.so

10 Copy the above rlogin to rsh also in the pam.d directory.

11. Fix .rhosts in user directory to rw only for user.

chmod 600 .rhosts

12. Change ownership for .rhosts

"chown user.user .rhosts"

13. From an existing RedHat 6.2 installation copy in.rexecd, in.rlogind, and in.rshd to /buildroot/usr/sbin/

Add whatever PVM application you need to the SuperRescue cd such as POVRAY.

With the above completed you may now burn a new cdrom and boot up your node. Upon bootup you should log in as root and startup the inet service. After that you can start up the network with "ipup".

Check and see what ip address you were given from the DHCP server. Log into the Host node as user and check that "rsh" is working by typing

Rsh remote_host ls

If all is working you should see a listing of the files on the remote host.

Check out http://www.netlib.org/pvm3/book/node23.html for other common startup problems.