Typing:

ifconfig eth0 down

while connected remotely to the machine in question, doh !

It is about time I connected the serial ports on my Linux and Win2k boxes together, so that I can recover from things like this without have to climb a ladder in the garage

Doh!

The other alternative is to UPS the box with a short delay, put the thing on an X10 socket and ensure that Wake On Lan is enabled.

Did Something very similar the other day. I was Dameware'd into a machine across town, to fix a problem with it's network connectivity. After changing a few settings to the correct one, I went to the network properties page (win2k) clicked on the net connection, disable. Very short pause, "FSCK!" as I realised what I had just done.

I had a bad habit of reconfiguring a Cisco router and doing it in the wrong order so that it shut down the interface I was connected through.

I've been working on my new server, and had a console on it set up to display logs on the old box. Few days later, I decided I needed to power down to tinker with the hardware, so I hit Ctrl-C, then typed in "poweroff". A few seconds later I saw my mistake as the console didn't display the expected shutdown sequence, but instead the broadcast message from my main server that the network was going down due to a shutdown command.

Wheeee.. The whole remote admin thing can get awfully confounding at times. It's easy to forget where you are and do something stupid. Of course, I realize what's going to happen at the precise moment my finger hits the enter button.

I try to avoid that with my bash prompts:

central = = = - - -
2398:drakino:~>

perilith = = = - - -
10023:drakino:~>

mavra = = = - - -
689:drakino:~>

nathan = = = - - -
1069:drakino:~>

There is a few others, but I figured 4 showed how I do it. I guess I might need to make each one unique.

Mine are always userid@hostname:last component of cwd:command number>

e.g.
shadow@johnstown:etc:1>

Here's one I did the other day (while logged into my webserver at home in London, from my laptop in Whistler):

root@peculiar:# killall sshd

Whoops. I had to explain to Jen how to reboot the box over the phone.

Well, at lest I don't feel so stupid now, seeing that the problem is not peculiar to me.

Nope. This kind of thing happens to everyone...

I hate when I do that kind of thing at work, because the servers I mess with are in a locked datacenter a few states away. I have to call a helpdesk, generate a trouble ticket and then call the datacenter with the ticket number for them to connect to that machine and restart the interface (with me explaining the commands step by step, depending on the skill level of the operator on duty...).

Fortunately, we also have remote console hardware so I do not have to do that unless I make a really spectacular mistake!!! The little ones I can fix directly.

My favorite:

admin> "The machine (sun box) keeps winding up at the OK firmware prompt."
me> "How do you know?"
admin> "We have the datacenter check the console for us."
me> "How do they know?"
admin> "They plug a keyboard and monitor into the sun box."

D'OH!

me> "You know, plugging in a keyboard sends a break on sun/solaris boxes. That alone will drop you to firmware."
admin> "...oh..."

At least they're not as bad as the Verio datacenter staff I've talked to. To get them to do anything to a server you're supposed to call them up quoting the machine number and then give a password. Nearly everytime I've called they didn't bother asking for the password. I've contemplated giving the machine number for Easyspace (big UK registrar that has their webserver colocated at Verio) a few times to see what would happen

One time I accidentally spilt some milk on my keyboard.