Victory ping.

“Whatever happened to the Legions of Doom server?” a coworker asked me as a technician swapped her computer.

I smiled a wicked smile. “Victory ping!” I then turned to my computer. “Ping pmprint02. Request timed out. Request timed out. Request timed out. Request timed out,” I read as the words scrolled onto my screen.

“Victory ping?” my boss–yes, my lunch ninja boss–came over and asked.

“I know that box,” the technician said. There’s a good reason he didn’t say “server.”

“Can I have a hard drive from it?” I asked, hopefully.

“NO!” he said, firmly.

“You can degauss it first,” I offered.

“I have a call in to a pyrotechnics engineer. We’re blowing that thing up. That’s that thing from over in Building 5, right?”

“Yes,” I said. I’d never seen the machine but I knew that was where it had been physically located.

“You realize that’s not a server at all. It’s a really old HP desktop that someone loaded Windows Server on.”

“A desktop? How’d they get RAID in it?”

He shrugged his shoulders. “All I know is it had no business being a server, and next thing we knew, they had everything under the sun running on it.”

It happens. Someone gets hold of a Windows CD and someone’s old workstation, and it starts out as a “development server,” but soon it’s a “staging server,” and after a week or so it’s production, but still running under some guy’s desk. No backups, no UPS, running on underpowered hardware, but golly, it sure is an emergency if anything ever goes wrong with it. I’ve even done it myself, once, but Barfy wasn’t a production machine.

I aloud wondered if that desktop-turned-server had RAID in it at all, or if I’d just read that between the lines when someone who wanted that server gone came to me and told me it was one drive failure away from never coming back. If it was a single-drive machine, it indeed was one drive failure away from never coming back. Or maybe it really did have an IDE RAID adapter in it, or maybe it had software RAID. Software RAID would have been fitting. Part of me hopes it was two consumer-grade Seagates from Best Buy in a software mirror. Striped would have been even worse, but then the server would have decommissioned itself within three years or so, so I’ll bet it wasn’t striped.

RAID or no RAID, I wanted it gone because it was running an out-of-support version of Windows and I didn’t want to pay for security patches for it anymore and the server team didn’t want to apply them anymore because it’s a royal pain, and the team that used the server always hassled them about rebooting the server and they were tired of the fights. As old as the machine was, we’d paid more in extended paid support for it than we ever “saved” by using a discarded desktop instead of buying a server. There’s smart cheap and there’s stupid cheap–I know about cheap, being Scottish and all–but this had definitely crossed the line into stupid cheap.

RAID turned out to be the key to get the server migrated to real server hardware running a supported version of Windows. I explained the server would decommission itself if a drive failed and then they’d be migrating during a production outage, rather than in a controlled fashion. A director or senior director said, “Well, we’ll just reboot it.” I said, “Please let me explain–if that array fails, there’s nothing left to boot. Our only longshot option would be to send it to Ontrack, pay $1,000 per drive to get it recovered, and it would take a couple of weeks.”

They got the server migrated in a week after I explained it that way.

It took 10 months for me to get that thing gone, and my predecessor had spent her entire career trying to get that thing gone. That’s why, from time to time, I open a command prompt and do the victory ping. Sometimes there are no sweeter words than “request timed out.”

If you found this post informative or helpful, please share it!
%d bloggers like this: