||Bastard Sysadm from Hell Manual
CRASHED - crash emulation daemon
Improves Unix to the standards of NT by crashing the system.
Actually better than the original because you can select when and from what
the system will be crashed. Various options can be used to select at what
time the system will be crashed. Using multiple options allows either of them
to do the job.
This daemon will usually be started from init(8).
Its name is pronounced crash-e-daemon.
emulate the "blue screen" visual interface, else simply the screen will
freeze with whatever is on it.
-e, --esp [DEVICE]
use /dev/esp or the given DEVICE to detect when a user will be maximally
-f, --kill-filesystem [DEVICE]
randomly write a few blocks of garbage to the root disk or to the disk named
-k, --key-presses [PRESSES]
only crash after any user has entered at least PRESSES key presses, so enough
work is lost.
don't crash while the operator is logged in, makes it easier to blame the
completely random timing, this overrides -e, -k, -o,
-t if they are before -r on the command line, else they are
added to -r.
used together with -k this resets the counter whenever a user saves
data, avoid useless crashes.
-t, --uptime [TIME]
after an specified amount of uptime(1), TIME in seconds.
-u, --user USER [USER]...
gives a list of users that must be logged in for crashing, together with
-e or -k only those users are monitored.
-b will only look right on an PC CGA/EGA/VGA/SVGA compatible console.
-e requires an BrainWaive compatible mind reading interface.
-f is only probabilistic, the file system may be dead or not.
This daemon has to be started by the system manager. This essential
functionality should really be in the Kernel - The nearest attempt
of this so far, the Linux jiffies overrun crash, can be emulated with
-t set to (2^31)/100.
Some users find this facility annoying.
The idea for this daemon literally fell out of the blue onto me.
But the behaviour it produces has been extensively tested by an
leading software firm.