A question about my servers hardware defunct...

[Kryssan]

I run a server from home, currently using Win Server 2008 x64.

Relevant stats:
TYAN S2895UA2NRF Dual 940 NVIDIA nForce Professional 2200 + AMD 8131 SSI EEB 3.5 Server Motherboard
AMD Opteron 265 Italy 1.8GHz Socket 940 Dual-Core Processor Model x2
CORSAIR XMS 1GB 184-Pin DDR SDRAM DDR 400 (PC 3200) x8
LSI LSI00005-F 64-bit, 133MHz PCI-X SATA II MegaRAID 300-8X Kit 8 Port SATA/300 128MB
8 250GB Internal Hard Drives configured into 2 RAID5s - One is the Application RAID the other is the Backup RAID
2 External 5.25" Hard Drive cage/carriages mounted in the CD bays for the OS drives, RAID1 250GB
Floppy Drive
CD/DVD Burner Drive


As you can see the hardware is getting a little old nowadays but as a server she's pretty rock solid and does just about everything I could want her to do.

However.


The one thing about ServerBox that's annoying me though is turning her off. It takes forever to get her to turn back on. The computer has never crashed on me (and I honestly wouldn't expect anything less from a Tyan board), and rebooting it from Windows doesn't seem to cause the issues. She'll stay up for months without complaint. But zOMG when she goes down (turned off in any way, either through Windows Shutdown or power failure etc) expect at least 30 mins to a hour to bring her back up. Anyone have thoughts on this? Everything from freezes during Windows bootup (load screen after Bios POST) to the computer just turning itself off after a POST check. The stuff is just completely random. I just have to keep trying and eventually I know after 2 or 3 attempts it'll boot into Windows. Never has it taken more than 3 attempts before it boots successfully. I can't find the problem anywhere, maybe it's just time to write off the MB and upgrade?

I just forsee so many problems with doing that though that it makes me edgy to try. And moreover I don't have the money to do it now anyway, getting tight with the command transfer and moving and stuffs.

[Zigma]

have you tried flashing your bios to the latest version?

[Zeith]

Sounds to me like it might be a problem with either your RAID or your RAID controller. I'd boot into the RAID BIOS and check it out, check out the health of the disks, make sure you don't have any repairs in progress. I've had the problem you're describing with various forms of RAID that use some type of parity block, as well as with non-RAIDed systems with failing hard disks. It's like there's a period where NTLDR hands things off to NTOSKRNL and there's no error checking on the data being read from the disk, so if it's reading garbage, Windows will inconsistently hang at the boot screen.

[Kryssan]

No, it's not a RAID or RAID controller issue as far as I can tell. It's all hardware raid as you can see, and the boot errors never occur during the LSI's post. This is however one of the reasons it takes so long to boot, as the LSI board has it's own post cycle to go through. However, this server has had a variety of OS's on it, and in the past when one of those was Linux, I was using a software RAID card and still had the same problems. I've cycled the disks out with spares, and the problems are still occurring.

Previously, I also had trouble getting this computer to turn off, but that was in a different case and a different setup. I think that was just a short in the 5v wire from the chassis. This MB/uP/Ram combination has seen 3 different cases now and has used everything from Ubuntu to WinServer 2000 to SUSE Enterprise to WinServer 2003 to WinServer 2008 it's running now. It's always had problems either with turning on or off, but that's really the only issues it's ever had. Its completely stable once its up and running. And really the only times I turn it off is for maintenance, cleaning, or moving things around. I guess I'm just a bit of a perfectionist in how I want it to work.

Now that I finally have a setup I'm happy with, I really don't want to rebuild it yet again. But, as I noted, and as you can see, the hardware is kindof aged now with exception to the LSI controller as that was a new addition I made to revamp it. Plus I'm kinda wiped for cash atm, but that's a problem that can be solved by time. Meh, if it comes down to it I guess I'll just suck it up and do it, I just don't want to, eh?

[Mafai]

Do you have another power supply you can swap in and try?

I had an issue before where once up and running my PC would be fine, but after a "cold boot" (overnight off), it would not boot up until I would try several times (as if it was warmed up from the first attempts). Swapping out my PSU did the trick.

[Kryssan]

As a true geek and computer nerd I have one of those nifty (and unfortunately stupid expensive) pci diagnostic chips that you plugin and boot up your computer and it'll tell you everything BIOS is doing and has about a dozen LEDs that let you know whether or not power is getting to the places it should.



Rest assured, it is not the power supply. I don't remember exactly but I think the PSU in ServerBox is a PCP&C, and I'd be completely shocked to see one of those go bad. Its been a while now since I popped the hood on ServerBox, so I don't remember. I've honestly been living with this problem for a while, I just thought I'd give another group of people a chance to see if someone had an out-of-the-box idea about it.



I guess I should have noted earlier that my diag chip doesn't report any problems with the POST check, either. That's why I'm a bit baffled on it. I'd say short, but, it does it outside the case too on an ESD surface, so meh.



I think eventually I'll just cough up the money and go through the process of rebuilding it with a new mb/uP/ram combo. I should be able to reuse everything else. Though, I might also change the fans out in the case too, because it can be a bit of a windtunnel if you're near it. Just using the stock Chenbro fans on it, should probably get some Scythes instead.



You know, honestly, just had another thought, once upon a time a few years ago this motherboard was watercooled and one of the HDD units had leaked, it was beneath the motherboard in one of those separated Lian Li cases so I don't think anything would have gotten on it, but maybe the root of all these problems stems from there. Hmmm. Oh well, to Newegg to wishlist some parts I guess... and wait for my paychecks.