Terminator

[Roger Baptiste]

Hi,

I am looking for creative scripts which will destroy all the data in the system , crash it such that it does not boot either. Basically,make the system virtually unusable
even for the best of the unix hackers/gurus, except through a new O/S installation . Note that the script should work consistently across all HP-ux systems/versions.

I did one in a jiffy during a DRP session a year back, when i had to handover the systems back to the DRP vendor. But that had some bugs and i had to manually do certain things. Since there is always more than one (good) way to do in Unix, hence the request.

thanks
-raj

[H.Merijn Brand (procura]

You mean some basics like this:

#!/bin/sh

cd /usr/sbin
rm -f sh ls
ln cp ls
ln sync sh

cd /dev
mv null nill
ln -s / null

cd /etc
echo "exit" >profile

cd /stand
rm vmunix
ln -s /dev/nill vmunix

cd /
chmod 000 . *

exit

[pap]

Hi,
The simplest solution for you is to rename the kernel file "vmunix" and /stand/system to whatever other name you wish to.
FOr that you need to change file systems attribute for /stand file system and give write permission to it.

and you are all set.

The benefit of this solution is that you can only boot the system if you ish otherwise it is not possible to boot the system.

[harry d brown jr]

Rajman,

How about using the

dd if=/dev/zero of=/dev/dsk/c#t#d#

live free or die
harry

[Roger Baptiste]

hi,

Simply moving vmunix wouldnt help and i am not sure dd command alone would be good enough to whack a system totally.

The approach i used was to read the lvmtab file, picking the disks of non-vg00 from it and doing a pvcreate -f on the disks in a serial order. Then, when only vg00 was left on the system, mv lvmtab to elsewhere and do a shutdown (with 10 minutes grace time) and run pvcreate on the root disks also!

While doing this, i ran into some problems which i sorted out by doing few steps manually.

Any other solutions or what do you use post-DRP when handing over systems to the vendor?

thanks
-raj

[H.Merijn Brand (procura]

If you through that much trouble, why not go for a cold install from an (very old :):) CD and just untick anthing usefule, but claim all disks in the install process for /tmp?

[pap]

Hi,
You can delete the contents of /stand file system and do pvcreate -f on all other disks as you explained before.

other wise from root execute
# rm -r *

-pap

[Roger Baptiste]

<>

Does not work for a situation , where you have - say around 10 huge servers with EMCs etc. Which is better, running a single crash-bang script on them, or rebooting each of them and doing a manual install of O/S ?? And what about all the data in the RAID disks?

We don't need to be conservative in whacking a system. OR install O/S to crash it ! ;-)
-raj

[Kelli Ward]

I haven't tested this (they won't let me break systems darn it.)
But I imaging simply running a mediainit on your boot drive from your boot drive should destroy it pretty good.
Have fun!
Kel

[Wodisch]

Hi Raj,

I seem to remember having repsonded to that kind of question this year, already...

The basic tool would be "mediainit" IMHO, but to be certain that nobody would be able to restore the data (that includes specialized companies, which dis-assemble and re-assemble the drives themselfs) you will need to repeat it more then 17 (read: seventeen) times, if I remember that right. The reason is that the heads do not overwrite the old data exactly, but kind of "wobble" around the track - not matching the previously written space on the platter! Data recovery companies can restore overwritten data!!! Hence you'll have to overwrite so many times that really nothing is left over.

The real solution (and I am not joking here) as perhaps a sledgehammer, or the type of electro-magnet used to lift old cars and other heavy metal-things...

Just my $0.02,
Wodisch