This probably isn't the best forum for this question, but I know several 'fritterers dabble in the Dark Side a bit, so I'm hoping somebody's got an answer for me.
I came by an older Intel server last week that has trouble running for any length of time. It's a dual proc. 850MHz PIII that's decked out with a gig of RAM, onboard LVD Ultra-SCSI, two PCI SCSI cards, and a slew of LVD SCSI drives. It's running NT-4 Server, but if I can get it to run consistently I'll likely load a copy of Linux (Ubuntu or Mandrake most likely) on it and use it as a home server for my Macs at home.
Most of the time it'll seem to fire up fine when you mash on the power button, but usually it'll get only part way through the boot sequence then silently start the boot sequence from scratch, like you'd hit the reset button. On the occasions that it does fully boot into NT4, it'll seem to run fine for a while, then BOOM, it restarts. No rhyme or reason as to when it happens, it just happens. The fans continue to run and the drives continue to spin, but it just reboots. Oh, and sometimes when it boots up it complains about one of the processors being offline, but usually it boots without complaint & shows both processors running just fine.
I'm thinking it's maybe a flaky power supply -- The voltage on one of the outputs dips slightly on occasion and the machine just stops -- but I could also see this having something to do with a bad processor or logic board. Or could it be something as simple as a dead BIOS battery on the logic board? Anybody have any advice for what to look for to troubleshoot this dog, or can you refer me to somewhere on the web that might be a good resource for a Mac-centric guy trying to figure out the trouble with an oddball machine like this? Or am I better off scavenging the useful parts from it & dumping the rest?
Thanks!
--
Dave
Usually this occures when the capacitors in the PSU get flaky, and are unable to hold a steady current and are constantly dipping. I recommend changing it ASAP due to you not wanting to have a brown-out across the system. Brown-outs can kill a machine faster than a spike in a voltage line.
I saw about 80% of the gadgets fry themselves in my house when we had a transformer in the alley out back blow half way, and put out only 70 volts or so. It killed the sony server LC 630 i had, as well as a 7200 and our original iMac Blueberry 333. I was lucky i didn't have any of my laptops plugged in
Needless to say, we have a surge protector added to the house via our Power company w/ brown-out protection.
About 7 months ago, we had another brown out, and it saves house, and it did it's job. it severed the power to the house, while my neighbor lost her tv, computer and SRS System.
I got lucky on that one...
Maaaaan, would that thing scream with Linux! *whistle*
How long has it been around, did it sit idle for a long while? How's the dust situation inside the case?
The first thing that I would do is a complete disassemble/reassemble, reseating all connectors, etc etc and get rid of any dust. Dust can be all sorts of trouble. Put cards, RAM etc when possible in the slots that they weren't in. Also, clean the edge connectors off with a pencil eraser.
During this dust removal, I'd make sure to blow air through the power supply and out. A lot of times when people are doing this sort of thing, they don't think to do the PS, and if it's clogged up, well then it would overheat, and then it'd be less efficient, and...
** Note: These may, but probably won't, solve things. But hey, they're all free to do, minus the can of compressed air you might need.
-- Macinjosh
The thing was used as a RIP (raster image processor) for the graphics department at the printing co. I used to work for. It had been in service for probably five years or so with no problems, then this issue came up and they replaced it with a newer/faster box. The guy responsible for maintenance there is a true neat freak; hardly a speck of dust inside the thing when you pull the side panel off. But I do know he's not brave enough to pull things apart too far, so there is likely dust hiding in the PS and elsewhere that he couldn't see from the open side panel. I had already gone through & reseated most of the components, with the exception of the processors, but hadn't gone so far as the eraser.
I think I'll go with Coius' recommendation & source a new PS for it. I have done a little shopping for one already, but sorting through all the options for PC power supplies is confusing -- another reason to like Macs! The PS that's in it is a Delta DPS-300JB... a Google search comes up with a lot of hits, and many of them are dealers willing to sell new & used like units, but don't know if it's better to stick with the same model or will a similar 300+watt ATX PS with the same plug configuration... Anybody have any thoughts?
Thats rather odd - we had a brownout here in the UK a few years ago and nothing was damaged. The voltage dropped from 240V to 70V. Both my PC's kept running (pritty cool that - also, many psu's will even run from a DC voltage source).
Any idea why nothing was damaged? Could it have been because all the equipment was designed for 240V?
I'd suggest simply going for another 300+ watt ATX PSU with the same plug configuration, but make sure it is a reputable brand. Cheap power supplies are cheap for a reason.
"when a brown-out occurs, a power supply will try to compensate from the lack of power, by upping the current in the power supply and ends up smoking itself out"
I have seen all the power transformers and stuff fry themselves within minutes. this occurs mostly in switching power supplies.
The same thing happens with an air-conditioner too.
Are those 850 CPUs Slot 1 or Socket 370? If they're Slot 1 I'd surely check if one has rattled a bit loose. Also, check if they've got removable VRMs, and if so check the connectors on them. If it's got a riser for the PCI slots, be sure that is fully seated too. A server isn't meant to be moved around a lot, so some stuff may have worked a bit loose during transport.
I'm almost sure there's a hood over the CPUs to direct air. Be sure it's there and on properly. Being a server, it should have a specific airflow pattern inside. Make sure no cables have been moved to block air from flowing where it should.
Well, I learnt something new tonight; never heard the terms "Slot 1" and "Socket 370" before. This machine uses Slot 1 processors -- they stand up from the logic board. I haven't figured out yet how they are released from their slots; there's some kind of latch keeping it in place. Each processor has its own fan (1" or so), and there's no hood. As for air movement, other than the fans on the processors, there is one fan on the PS, two on the drive rack, and one that pushes air out the front of the case.
VRM... That's a voltage regulator module, correct? I'm not very familiar with Dark Side hardware, so what do they typically look like?
On each end of the processor module there is a notch that is held down by a tab/clip that sticks up from the mobo.
The VRMs looks much like the ones you'd find in a Beige G3, if they are separate. If they are built in to the mobo, they just look like a cluster of large capacitors.
Check the direction of airflow on those fans. Generally a server has the fans suck in the front and blow out the back.
I agree. I had some restart issues while connecting a customer's hard disk to one of my computers for a virus scan. With everything plugged in, it restarted in about 30 seconds. After disconnecting the power from the optical drives, it worked. While it had an Athlon 64 3700, 1GB RAM, 2 optical drives and a single IDE hard disk, I didn't expect it to do that. This was supposed to be a 450 watt ATX PSU.