Epoch Fail

Thank you xkcd, for coming to the fail-party...

The largest [and fastest?] re-boot

[ The Register calls it Reno 911 ]

Reno 911: World's largest reboot underway

The sound of 10,000 OSes clapping

By Ashlee Vance in Reno

Published Tuesday 13th November 2007 22:03 GMT

You were up all night writing those last lines of code to ensure mega-demonstration success. And this code is a real pain in the ass to deal with because it has to spread across a 72-processor cluster. But, with a bit of perseverance, you nail it, pop open a beer and wait for the glories to follow when show attendees see your genius on the conference floor.

Cut to the actual demo, and you're screwed to all hell when the power goes out.

Yes, Supercomputing 07 went through what could be the world's largest, collective reboot of all time. At about noon today, the immense amount of hardware here in Reno killed the power system and then killed it again a few seconds later. Lights went out. The machines stopped humming. And a few claps scattered throughout the crowd - a machismo kind of thing no doubt: my cluster is too big for your power station.

As the power returned, machines of all shapes and sizes rebooted. There were Windows log-in screens everywhere. Versions of Linux going through their routines. There were also the frantic types pulling off their chassis casings when their systems failed to come back to life.

Er, this was a kick ass protein folding demo

Every year, the machines present at Supercomputing combine to form the equivalent of one of the world's largest machines. Some of the machines here make an HP Superdome look like a tinkertoy.

Breath deep. It'll all be okay

Sanity seemed to hold though as these folks temporarily lost their precious machines, which as an encouraging sign. Now, the electricity collapse can be enjoyed as a point of pride. ®

Update:

As we understand it, the power outage occurred after some renegade plugged in a four-socket Itanium server using the new Montvale chips. (Sorry, couldn't resist.)