Can I bring a Win2003 server back up on different hardware using the same disks?

Question

I have a Windows 2003 server that has hung twice in the last two weeks - the second time after less than 2 hours up. There is nothing at all in the Event Viewer to indicate any failures. For a freeze like that with no disk errors, I suspect some kind of failure in my server hardware.

My server is a semi-generic 1U Supermicro with an Adaptec RAID 1 controller. I'm running two SATA 500gb drives in RAID 1.

I have an older but functional Dell server box that I'm not using right now; however I don't believe it has the same Adaptec RAID controller. (I'm actually not sure, but for the moment I'm assuming not.) I'm trying to figure out whether I can take a disk out of my failing hardware and bring the Win 2003 server back up on the Dell hardware.

This breaks into two sub questions:

1) If I take a disk that was part of a RAID 1 set, is there a way to boot that on the Dell server assuming the Dell doesn't have RAID?

2) How picky is Windows Server about the hardware? Will it detect that things have change and boot successfully on different hardware?

And is there anything else I should be aware of?

Answer 1

The main problem your going to have is with the disk controller, if your new hardware has a different disk controller it is more than likely that windows will blue screen when you attempt to boot.

If your disks are configured in a RAID array then your not going to be able to move them to new hardware, as the configuration data is all stored on your RAID controller.

All in all, its probably a bad idea, you'd be better of using some sort of imaging software to take an image of your disks and move them to new hardware.

Answer 2

This is a dangerous move on your part, but if you're not concerned about loosing your data, read on...

1. That's the curse of hardware RAID. Once you have drives in use with it, you'll want to stick to whatever hardware card is attached to it. Disconnecting the drives and shoving them in JBOD-style is a long shot and may or may not work, depending on how the RAID controller handles the disk layout. Some controllers "write" info into the disks and this can basically ruin your idea from the start, while others store everything in battery-backed CMOS and really do treat it as a pure mirror.

2. How close are the two systems related? Do they have (roughly) the same motherboard chipsets? Windows is fairly tolerant of being moved to different video cards, NICs, drive adaptors, etc. but it tends to not like being swapped to a motherboard with a completely different Northbridge/Southbridge.

My answer is a cautionary "Yes", provided that (a) your RAID controller isn't writing funky things to disk, and (b) you look at the differences in drivers and you're prepared to get it working by providing the missing driver software. Silly as it sounds, try installing the drivers before you move the disks over. Initially booting into "Safe Mode" might be your best option to try this. Keep in mind that once you break that mirror, getting the drives to sync might lead to some data loss situations (I have seen confused RAID controllers sync the "old" drive to the "new" drive when you rejoin a mirror, leading to data loss).

My personal recommendation?

I would arrange for some downtime, then proceed to:

• Vacuum the board and components with a data vac, or if you can't do this, use compressed gas (not recommended but useful as a last resort) to get the boards clean of dust bunnies
• Check the heat sinks on the CPUs and other components for clogged dust bunnies. Those fans do an excellent job of sucking in mounds of crap over time, and the heatsinks make for great dust traps. Modern "CPU smokestacks" on some smaller cases make the problem worse, turning your computer into an expensive vacuum cleaner.
• Using a flashlight, inspect the motherboard for damage, swollen or blown caps, or other issues. Cooked, blown, or clipped components will allow the computer to operate in a substandard fashion for years.
• Using proper ESD precautions, re-seat all of your slotted cards and RAM. Over time, the components will expand and cool due to thermal stresses, and on rare occasion, poor or shoddy manufacuring tolerances in the CPU case will lead to warping that will literally pull cards out of the slots. I've seen a PCI network card "pull" itself out in a cold manufacturing environment, and I've seen RAM that won't quite seat up and eventually wiggles its way out of the socket.
• Ensure that the power supplies are also cleaned/blown, and that any fans attached to the power supplies are working, or if it's a passive system, that air is flowing through them before powering the unit off.
• If this is a low-end or whitebox unit that uses aftermarket parts, you may wish to look at re-greasing the CPU and heatsink. I've seen old thermal paste develop hotspots under the socket, leading to intermittent freezes on the CPU. Once the new paste was applied, the issue went away. Be careful about this. If it's mid-to-high grade equipment, the paste is often a specific "pad of goop" that is supplied by the vendor and you won't be able to perform this procedure. If in doubt, don't do this.

Answer 3

You can try slapping the disks into the Dell machine, but if your goal is a stable, working computer you're better off installing from scratch on the Dell machine, or ditching the hardward RAID controller on the SuperMicro machine and rebuilding from scratch with a JBOD configuration and using Windows software RAID.

I know that's not what you want to hear, but it's the right thing to do.

I've tried to do that with at least one Adaptec SATA RAID controller and gotten BOOT_DEVICE_INACCESSIBLE STOP messages. I can't remember which model to tell you for sure.

Rebuild the box-- you want it to be stable.