"Please don't press this button again"

[coyoteden.livejournal.com]

from LJ Dev: "Another customer in the facility accidentally pressed the EPO button, then depressed it, replaced the protective case, and left the building."

Hhh...huh... HOW DO YOU "ACCIDENTALLY" PRESS A COVERED BIG RED BUTTON!

Accidentally? Bullshit. That was sabotage.

Comments

[ozdeadmeat.livejournal.com]

Bloody Terrorists!

Ah, yes, The Big Explanation...

[principia.livejournal.com]

I'm sorry, but Internap needs to send someone with a middle name of 'The' to go visit that particular customer.

Re: Ah, yes, The Big Explanation...

[jecook]

Probably already happened, and the person responsible likly sacked.

It may even be possible that internap kicked them out as a customer, although that's not likely.

Re: Ah, yes, The Big Explanation...

[boredevilperson.livejournal.com]

Like . . . Kermit The Frog?

Re: Ah, yes, The Big Explanation...

[110billion.livejournal.com]

I was thinking Ivan the Terrible

Re: Ah, yes, The Big Explanation...

[110billion.livejournal.com]

even better :D

Re: Ah, yes, The Big Explanation...

[principia.livejournal.com]

More along the lines of 'Zippy "The Weasel" Bataglia'. :)

[(Anonymous)]

Cleaning ladies have done it twice in our colo.

Do not underestimate the power of human stupidity, especially of those making minimum wage.

[jahbulon.livejournal.com]

What does this big red EPO button do? It sounds explodey.

[torkell]

It's the "kill all power to the entire building and take out the UPS system for good measure" button.

Not one to be pushed lightly.

[erratic0101.livejournal.com]

I dont get it. How can an EPO switch kill power that comes from batteries??

[jecook]

What the SCRAM switch (or EPO, or emergency power off button) does is open the breakers at the power distribution bays that feed each row or racks. Commonly the building's UPS is tied into the distribution just before the EPO breakers. This way, if something truely bad happens ( like a large truck comes through the wall at high speed and take out a row's power distribution frame) you won't have live electricity anywhere in the room.

http://www.faqs.org/docs/jargon/S/scram-switch.html

[erratic0101.livejournal.com]

oh! it doesn't just turn the power OFF, it CUTS it off. Sheesh.

[jecook]

Yep. It can also initiate truely expensive things, like a halon dump, or other fun and exciting things.

[jahbulon.livejournal.com]

ah, the doomsday button. I thought it was only a legend.

[torkell]

Apparently not so, although the closest I've seen to it in real life is the small unguarded button on the front of the UPS at my old school labeled 'O'. No prizes for guessing what it does...

This may be a stupid question, but -

[the-s-guy.livejournal.com]

Why was this button in an area that the CUSTOMERS had access to?

Re: This may be a stupid question, but -

[jecook]

Because you would want something like that in an enviromentally controlled room with several million dollars of equipment. Trust me.

The "customer" in this case was another person or company like Livejournal or some such that has equipment co-located there.

Usually, the people that are allowed access are either escourted, or know enough about said rooms to not push the button except in case of a true emergency.

The Co-lo's I've been in were all controlled access facilities. I'm pretty certain that Internap has a clue as to the company that did it. One of those facilities used biometric Identification, a PIN, AND my badge. Other's just has a badge. If I was one one of the companies non-owned sites, I was usually escorted in/out and someone watched as I did my work.

Re: This may be a stupid question, but -

[dr-atheist.livejournal.com]

The button is a health and safety requirement, it's for fire crews to power off the building in the event of an emergency.

Take a hint from ACPI PCs

[coyoteden.livejournal.com]

IMO, buttons like that should have to be held down for at least two seconds before they do whatever drastic thing they are suposed to do. If there's a fire or someone is being electrocuted, two seconds is not going to make all that much of a difference, but it sure will prevent someone from killing an entire machine room by bumping it with a box they are carrying.

Your PC's power button probably requires a long press to kill power immediately. Why can't the power button for hundreds of machines work the same way?

Re: Take a hint from ACPI PCs

[thecrazyfinn.livejournal.com]

Because when you need it, you need it right now.

As what happened when an idjit dropped the Silver Spanner in a Bell CO in Toronto in '99, taking out 151 Front St, and thus most of teh 'net in Ontario back in '99. It might have been savbed if the power had been cut immediately, as it was, a 12 hour outage occured.

Re: Take a hint from ACPI PCs

[the-s-guy.livejournal.com]

Hmm, I can see one second being a good compromise. I'd also expect it to be recessed behind a surrounding fence of minature steel rods, so that it had to be jabbed by a finger or bent elbow rather than triggering from backing into it or thwapping it with the side of a box.

A better alternative - two such buttons next to each other, where a thumb and finger on the same hand could press them both. Unless both are pressed at the same time, no fireworks. And make them slow-release pushbuttons, so if they're triggered in series within a 2-second period, it still counts. You can still trigger a no-delay response by jamming both buttons down together.

Then put them on the wall in a cleared area with no other equipment or devices for ten feet in all directions. Paint the wall in decreasing bullseye rings around the buttons. Paint the floor, ceiling, and the edges of the wall area with black-and-yellow-stripe danger colors. Raise the floor an inch (with a ramp-slope, not a step) and rig the raised section to boom noisily when someone puts weight on it. Install one of those college fire-alarm devices that sprays purple or ultraviolet ink onto the hand of the person triggering it.

Rig a little webcam to a fire-hardened microstorage device and UPS, and point it at the buttons. Configure it to record the previous and subsequent 30 seconds surrounding the triggering of the buttons. Mirror the feed via shielded fiber-optic to a UPSed monitor and securitycam recorder outside the room, and from there via a wireless connection to another building or satellite.

OK, maybe overkill a bit. Or would that include multiple cameras pointing at the buttons and the rest of the co-lo area, all with UPSes and feeding by satellite to three different security firms in the next state over?

"We know who dunnit! Er... the boss tripped over and faceplanted the buttons. But at least we have it on video!"

Re: Take a hint from ACPI PCs

[microchip.livejournal.com]

Hell, stick it behind easy-shatter glass type stuff, the type on trains (at least over here!) for emergency stop buttons... if you *need* it hitting, you just go straight through with your elbow or something similar, and it's done. Provides *some* protection against idiots, and requires virtually no extra time delay, as you can punch through it without damaging your hand. Aside from it maybe aching a bit later.

Re: Take a hint from ACPI PCs

[microchip.livejournal.com]

... Hang on... it had a protective case on the damn thing...

That customer needs beating. Severely. And then some.

Re: Take a hint from ACPI PCs

[mg4h.livejournal.com]

It says the button was pressed, then depressed. That indicates to me that there's a difference - just pressing it a little or for a second wouldn't do anything. The fact that it was also depressed was most likely the few-second safety that this numbnut bypassed to turn everything off.

I still cannot fathom someone doing that, unless they're a complete maroon. Gah.

Re: Take a hint from ACPI PCs

[coyoteden.livejournal.com]

I think depressed in this case means "de-pressed" as in it was pulled back out. Some of those EPO buttons are like fire alarms or the stop buttons in elevators in that when you press them, they stay in and have to be pulled back out to reset them.