How To Make a Password Cracking Cluster

[djscribble]

Originally posted here by djscribble
IronGeek: when you say that if you aren't an admin to copy the sam and then use saminside are you talking about using linux w/ ntfs read ability?? i thought the file was "protected" by windows so that you couldn't do that...

iron geek: another thought on the sam and system file -- you can get it from c:/winnt/repair or c:/windows/repair so i guess we are both kinda right...

[NeuTron]

Originally posted here by johnnymier
First of all, Great tutorial NeuTron!

I setup the cluster in 4 P4 boxes 128 ram each. Passwd file had 3 accounts with 6-8 characters (alphanumeric). The first password took about 35 seconds. I left the cluster running for about 50 minutes and couldn't get the other two accounts. Is that normal behavior?? After that time I had to leave the computer lab at uni because some other class needed to use those boxes. I am keen to try it again next week with maybe two more machines in the cluster. I will let you all know of my results.

cheers

-j

Johnnymier, my cluster didn't have enough processing power to crack all my passwords either. Only 2 P4's and 2 P3's. Thats why I want EVERYBODY who tries this to post the size of the cluster and the time it took to crack their list. Also mention the compexity and number of the passwords. This way we can benchmark how much processing power it takes to crack certain types and lengths of PWs. Thanks to everyone who has replies so far, it has motivated me to write more tuts like this one.

[sec_ware]

Hi.

Cisilia looks like a sophisticated project. Just out of curiosity:
Has anyone looked at the scalability?
Has anyone tried to use it on opteron/itaniumII-processors (I got a
segmentation fault doing ./configure, make. However, I did not try
to change any cisilia-default flags)?


I'd like to add a reference to a "poor man"-version:
the "distributed john the ripper"-project [1], which
is based on john the ripper [2]. Maybe some of you
already have looked at it, since it was mentioned in [3].


Description: The way it works is different than cisilia's -
as much as I understood of the latter. The server "djohnd"
simply splits the password-file, here called passwd,
into smaller pieces and "distributes" the task of cracking
onto several nodes (in a client-server manner). The clients themselves
use "john the ripper" for that purpose.
I guess, after that description you might know, why I call it
a "poor man"-version .


Application: For administrators of large systems with many users,
this might be an interesting option. Use idle-time of your machines
to check for weak passwords, coordinated by "djohnd".


Security issue: djohnd is in its current version highly sensible
to DOS attacks. So, and not only because of that,
use it in a trusted environment only.
Hmm, or not at all




[1] http://freshmeat.net/projects/djohn/
[2] http://www.openwall.com/john/
[3] http://www.antionline.com/showthread...r&pagenumber=3

[Jareds411]

Does anyone know of a program like cisilia to crack other types of things, or does anyone know of a clustering system that splits up individual processes?

[AxessTerminated]

Jareds411. Cisilia is open source. You could modify the source, adding algorithms yourself. You could even start a project to do so on something like SourceForge. I think it would be a great idea if Cisilia had the capabilities to crack passwords, encryption schemes, EFS, and even RSA and PGP style encryption.

A_T

[Jareds411]

I'm sure I could do that pretty easily, seeing how I'm 15 and barely know any C++.

Back on topic, when I'm running clusterknoppix, it won't let me write to the harddrive, only read. Does anyone know how to change that?

[mgomespt]

Has anyone successfully changed the alphabet being tested by cisilia??? by default it only tests uppercase chars and numbers, i wanted it to include lowercase also...i've tried building a alphabet file like in cisilia man page, but it seems upper and lowercase get recognized only as being uppercase...(i see this on with verbose on (-V) in cisilia)...as far as i know windows XP passwords can get both upper and lowercase (of course they can, my login password uses both!)
can anyone help me?

[Mad_Whitehat]

hmm i tried your technique in my computer class, but nodes are not automaticly added, it could be i did something wrong, or i am using a different version of clusterknoppix, i downloaded 3.6.

Which version did you download and got it working that nice :S?

[Mad_Whitehat]

Originally posted here by mgomespt
Has anyone successfully changed the alphabet being tested by cisilia??? by default it only tests uppercase chars and numbers, i wanted it to include lowercase also...i've tried building a alphabet file like in cisilia man page, but it seems upper and lowercase get recognized only as being uppercase...(i see this on with verbose on (-V) in cisilia)...as far as i know windows XP passwords can get both upper and lowercase (of course they can, my login password uses both!)
can anyone help me?

probably it tries to crack LanMan hashes, these are the easiest cause they can only be uppercase chars, while NTLM can be Upper and Lower case chars. I think this is the problem you are discribing

[NeuTron]

Originally posted here by Mad_Whitehat
hmm i tried your technique in my computer class, but nodes are not automaticly added, it could be i did something wrong, or i am using a different version of clusterknoppix, i downloaded 3.6.

Which version did you download and got it working that nice :S?

Do you have a DHCP server available? Are the nodes x86 architecture? Oh, and 3.6 is the most recent.