Dying Power Supplies

[Taddeusz]

Gotta love it when a power supply dies and takes some other stuff with it. Here at work we have a couple CAD workstations made by TriStar. It is in an Antec case with an Antec power supply. The user had been complaining about it not turning on occasionally. The first thing I thought was bad capacitors in the power supply as I've had ALL of my Antec power supplies die that way. I popped the power supply open to look and it seemed fine. The voltages in the BIOS looked good.

That was probably about a month ago. Fast forward to today. Some time over the weekend the power supply, I'm quite sure, let out the magic smoke. He came in this morning expecting to move his mouse and log in. Instead the computer was off and no amount of power button pressing or cable fiddling brought it to life.

I came in about 15min after the call. Had to wait 2 hours so I could go to the nearest Best Buy for a power supply. I could have gotten some Chinese brand that I'd never heard of an hour earlier but I probably would have trusted it about as much as I currently trust Antec.

Got the new power supply in and the system powers up fine. The first thing I notice is that the CPU fan is not spinning. I go into the BIOS and notice that none of the system monitoring is functioning. No voltages, fan speeds or temperatures. Then I try to boot. Nothing. It says there are no boot devices available. I pull the drive out and try to turn it on again and the drive isn't even spinning up. The drive is an 80GB Seagate 7200.9. I remember I have a similar drive at home I can try swapping the controller board with. No go. Not only that but there are physically charred components on the controller board and it smells nice and toasty. Definitely an exit point of the magic smoke.

So a new computer is ordered (not from TriStar) since nothing in this system can be trusted to be good. The hard drive is also on its way to OnTrack to see what data, if any, can be recovered from the drive. Judging by the charred components it appears that everything from the power input to the motor controller, if not the motor itself, is fried. They are probably going to have to remove the platters to recover any data. If they can recover any data the money will be well worth the man hours that were put into creating the data

[panteragstk]

That sucks. The only psu I've ever had die on my was an Antec. I'm sticking with corsair from now on. Good luck on the data recovery.

[Taddeusz]

Yea, I've had no luck with Antec power supplies. All three of them that I've ever owned has died of the capacitor plague. Antec makes nice cases but I have zero faith in their power supplies.

[Thomas Wischgol]

I agree. The Antec power supplies definitely do not last very long. Actually the only brand here at home that died on me so far. On a positive note, if your hard drive died due to the power, the platters are probably still intact which means that the data should be recoverable.

[jptheripper]

wait, sorry to be a knob, but you have data on a hard drive worth the thousands of dollars of the cost of data recovery, but it wasnt worth the 100 bucks and 10 minutes to at a minimum mirror to a backup usb drive?

really?

[Taddeusz]

We actually do keep backups just not of individual workstations. All user document folders are located on a server and backed up nightly. In this particular case the user had checked out a CAD drawing to work on and made significant changes but had not checked it back in since it was still a work in progress. It could possible be prudent to do back up his local repository when he has critical stuff like this out.

[jptheripper]

Nor do we, but as a developer not willing (or able to remember ) what i did yesterday, i took it on myself to copy my local development CVS folder to a network storage drive with a nightly scheduled task. That way i had a backup whether or not i checked it in, of at least a week or so. It at least provides a buffer against catastrophic failure, or catastrophic stupidity.

[panteragstk]

Speaking of catastrophic stupidity, I was an admin for a state run organization that only backed up servers. ALL data was stored locally on the workstations and NOTHING was ever stored on the servers. Needless to say when a hard drive crashed user literally lost decades worth of data. I didn't work for them too long. I can't stand when companies are too cheap to actually keep their business running smoothly. They were still running Pentium II servers in 2005.

[Taddeusz]

Ouch. That is pretty bad. If they didn't store anything on the servers what were they used for? How did people share information without having to email it back and forth? Kind of sounds like the data distribution was probably a result of the work environment and they have an unwritten policy of secrecy unless others need the information to do their work. Job security through job obscurity.

[jptheripper]

I'm the DBA for our GIS (mapping) group. High end software, ~= AutoCAD from a load prespective. We currently support 20 concurrent versioned editors, running on an Oracle database, on a Solaris 280R (dual 900Mhz processors no less). I would love a Pentium II

Yes this is our production environment.

Originally Posted by panteragstk

Speaking of catastrophic stupidity, I was an admin for a state run organization that only backed up servers. ALL data was stored locally on the workstations and NOTHING was ever stored on the servers. Needless to say when a hard drive crashed user literally lost decades worth of data. I didn't work for them too long. I can't stand when companies are too cheap to actually keep their business running smoothly. They were still running Pentium II servers in 2005.

[panteragstk]

Originally Posted by Taddeusz

Ouch. That is pretty bad. If they didn't store anything on the servers what were they used for? How did people share information without having to email it back and forth? Kind of sounds like the data distribution was probably a result of the work environment and they have an unwritten policy of secrecy unless others need the information to do their work. Job security through job obscurity.

They kept the records on their hard drives. All the servers were for was connecting to the other parts of the state network to process things like medicade and other stuff. I had users with 4gb pst files because they kept ALL of their emails. All of them. They actually had com and queue servers because there wasn't a way for one server to do everything. Novell netware was used for all functions on pc's running NT 4 to xp and everything in between. It was awesome(ly bad.) I actually got in trouble for not removing spyware (which the state had no software in place to do so) when it came from the ghost image cd's we used FOR EVERY COMPUTER. Needless to say I don't miss that job. To make matters worse I was a 21 year old guy in an environment with nothing but ~40 year old women. I was not liked for being smart and young (and male). My boss actually asked me when I was getting married because married men were easier to control. That and she wished she could hire nothing but married women. I quit after that (and HR said just to wait for her to retire in 3 years).

/rant over.

Originally Posted by jptheripper

I'm the DBA for our GIS (mapping) group. High end software, ~= AutoCAD from a load prespective. We currently support 20 concurrent versioned editors, running on an Oracle database, on a Solaris 280R (dual 900Mhz processors no less). I would love a Pentium II

Yes this is our production environment.

Wow. I'm guessing replacement parts are very easy to come by.

[jptheripper]

Originally Posted by panteragstk

Wow. I'm guessing replacement parts are very easy to come by.

Fortunately maintenance cost is higher than replacement cost, so we finally have a modern server replacement in the queue. Unfortunately time it takes to move applications off a server that has been in service since 2002 is measured in months to year+

[panteragstk]

Originally Posted by jptheripper

Fortunately maintenance cost is higher than replacement cost, so we finally have a modern server replacement in the queue. Unfortunately time it takes to move applications off a server that has been in service since 2002 is measured in months to year+

Man, that sucks. That must be one heck of a lot of applications.

[jptheripper]

just a few. The main problem is people. 100+ servers, 1 SA. 80+ databases, 2 DBAs. not enough network guys. Backlog of work months long. Just getting a new server commissioned, after delivery, is 6+ months.

When your budget comes from property taxes, and property values get cut in half, the services you can deliver take a huge hit.

[stevech]

My employer preached personal and IT security to the max. Holler, yell, threaten. Forced full-disk-encryption on all of us. Disable Admin rights on PCs. This is a Fortune 500 company.

One day, burglars broke a window, entered the HR department, stole the desktop PCs, on which unencrypted HR files and financial records were stored. No servers. Not a backup issue, but a sad example of moronic large-company management that doesn't know what color socks they have on.

[panteragstk]

Originally Posted by stevech

Not a backup issue, but a sad example of moronic large-company management that doesn't know what color socks they have on.

Very sad. Of course they both LOOKED blue...