In some cases it can be impossible to restart your server or just shut it down. The reasons could be many, but most times it concerns I/O blocking that could be caused by a faulty device. If you’re having the server next to you, you’ll just hit the reset button and disconnect the faulty drive. In cases when the server is not in your reach you can send a message to the kernel to HARD reboot. All commands are of course done with a root account.
echo 1 > /proc/sys/kernel/sysrq
echo b > /proc/sysrq-trigger
Keep in mind this is the same as pressing the RESET button on a an active machine. This can corrupt data if unlucky. Dropping 1 into sysrq means that you want to access a system request. This is normally set to 0. And the trigger ‘b’ means reboot system. It’s not good practice, but it helped me to get a system back up again after hung processes eating all the IO.