Here’s a rather brutal tip that can come in handy when you administer remote servers: should you discover a process that’s gone to a D+ state – Uninterruptible Sleep – because of a blocking I/O call, how would you go about instructing the server to reboot without having to wait for all processes to terminate, since we all know that’s never gonna happen ? Even a kill -9 won’t terminate the bugger, so here’s how to bash the server’s brains for good:

echo 1 > /proc/sys/kernel/sysrq
echo b > /proc/sysrq-trigger

Needless to say you have to think twice before applying this rude trick, since it’s pretty much like yanking out the power cable (without the stress on the electronic parts). You may lose data by doing this, so make sure you’ve shut down as many daemons as possible, especially those which access block devices or sockets and files.