Vista Boot (Drive 0) again! Recoverable?

#1
After MUCH searching, I finally found some useful posts here on this topic, but they didn't answer anything (thanks to those that posted!)

I initially set up my Gigabyte P35-DQ6 system with 2 150GB RAID 1 drives (Vista and XP partitions), and 2 80GB drives (intended for Ubuntu) on the Intel controller and installed Vista and XP just fine.

I went to add two 500GB drives on the Gigabyte (JMicro) controller only to find that the driver for it would make vista BSOD on boot. I moved the drives to the Intel controller, and Vista wouldn't boot at all (I figured it might be "drive 0 related" reordered the drives (physically), but got no change (it seems Vista ALWAYS orders RAID after non-RAID volumes, even eSATA.) I discovered if I power on the eSATA drive AFTER vista is booted, it's fine.

Eventually, though, one problem (power, boot, BSODs or whatever) caused a RAID failure (I got one drive (a WD Raptor) reporting error, and during the rebuild from the OTHER Raptor, the SECOND reported failure and I lost the array. I strongly doubt both Raptors failed.

I reinstalled Vista, and after discovering VMWare realized I only need the Vista partition, so I took the 2x80 and 2x150 GB drives and threw them in a RAID10 array for the 25% reliability gain in case one of those drives really was corrupting the array. This time it DIDN'T BSOD with the Gigabyte (JMicro) driver, so I could read off that drive finally. Not until AFTER I got Vista reinstalled did I discover the boot files got shoved to what it now thinks as disk 0,0, the 1st partition on the data drive on the second controller. This of course as someone else mentioned affects the complete system backup (not a big deal it's just one partition lost in this case), but I fear that it defeats the point of the redundancy in the RAID setup.

My question is this: If the data drive were to fail, the one with the /boot data on it, can the system be restored with just the RAID Array with the DVD's "restore boot" option?

Or could the system be restored if I replaced that failed drive with another drive with the DVD's "restore boot" option?

And if, in a catastrophic event of a PSU blowout or something and all the drives get taken out, if I replace the array, AND the stand-alone data drive (with different partition configuration as long as a 0,0 exists on it) would a system complete restore from my backups work as long as that drive is there, or am I set up for a non-recoverable disaster in any case making the RAID array nothing more than wasted space?

BTW: the image generator used for forum registration is the nastiest I've seen, it took 5 attempts before I got a letter set I could almost make out :smile:
 

mqudsi

Mostly Harmless
Staff member
#2
Welcome to NeoSmart Technologies, Tundera! :smile:

I'll take that as a complement :grinning: - spam is everywhere and I'm pretty darn impressed with this new CAPTCHA. It's a PITA if you get a bad set, but the benefit is once you've registered there's no doubt you're a real human :tongueout:

As for your question....

1) Both would work. You can recover the entire system from the RAID mirror or with any other drive along with the Vista DVD's "fix startup" feature.
2) Should work. And if it doesn't boot from the restored images, just run the "Startup repair" feature to get around that :smile:

{BTW, this is off-topic, but I'm planning on getting the DQ6 w/ 2x500GB Samsung Drives, 2GB memory (GeIL), and an E6750. Am I wrong in thinking its a great board?}
 
#3
Well, all of those scrambled letter codes are difficult to read, but I've never seen one that overlaps a bunch of letters and so many odd angles :smile: I think it may help to have those red and blue mylar lensed 3D glasses :smile:

I'm glad to hear it should work to use the repair...I had heard somewhere that it wouldn't, and some nasty process about going to console recovery mode in another Vista installation and copying out the boot directory and BCD file, and another post/article somewhere saying that it's bound to the disk ID and so even copying it wouldn't work, and somehow the recovery tool would miss it if it's not on the same disk originally...it's was a bit of a scare thinking that a 4-drive boot array could still be rendered useless by a single data drive failing :smile: Vista REALLY has issues with handling RAID arrays with it's boot sector!


As for the DQ6 from what I've seen of it, it's an amazing board. I previously got the Asus Striker Extreme, and after that disaster, I'm not sure I'd ever buy another Asus product despite liking them previously for many years. It was RMA'd within days. (Yeah I know Gigabyte is sort of part of Asus and sort of not, now, but clearly their design teams have a better clue :smile: )

The Striker, while featureful, had a LOT of odd quirks and problems, and that specific one had an issue where it wasn't sending the power-ok signal back to the PSU, so the fans would start to spin up, and then everything would shut down before POST, then randomly it would work for a boot or 10, and then not boot again. And it ran *HOT*....the moment it went on I felt hair-dryer heat coming from the CPU/northbridge heatsink fan.

The P35-DQ6 (despite that nasty RAID crash) has proved wonderful. The overall feel of it is very study, the chokes are HUGE, and all the capacitors are solid and SMD mounted. They're not joking, it's really quite industrial...even thicker than my 3-layer Opteron workstation board :smile: And it runs VERY cool, actually cooler than the Opteron, amazingly, and only consumes slightly more power (which is remarkable considerring that energy usage includes running 4 EXTRA HDDs, 4 1GB DDR2-800 DIMMS versus 2 512MB DDR1 -400 DIMMS, and it's a Q6600 CPU....the board's definitely burning less power, that's for sure. Other than the fact that getting to all the SATA plugs when you have a jumbo (8800GTS) VGA can be a bit hairy, I've no complaints so far. It's cool (with hot components), it's stable, and it's well built!

As for the array crash, there's a few possibilities for what caused it (and I doubt it was the board)

It could be:
1) A real bad sector on a Raptor that I haven't hit again yet
2) Vista 64's BSODS finally did odd things to the array
3) (Very likely) The weight of my motherboard power cable kept pulling it away from the motherboard where after a way it wasn't making contact...I've hung it from a support on the chassis now)
4) (Even more likely) Vista's "Sleep" mode (avoid this like the plague if using RAID, it does odd driver reloads in the process, and I imagine for RAID, bad things can happen in that process :smile:

I think you'll like the DQ6 board :smile:
 

mqudsi

Mostly Harmless
Staff member
#4
Hey, thanks for the advice :smile:

To check if your issue is caused by the Raptor, just run chkdsk /r
But I'd have to agree with your synopsis, Vista's sleep is the pits. We actually posted about it here, you won't believe the number of comments from people experiencing the problem! It's like sleep problems are the norm and it's only the rare few that can it get it to obey!
 
#5
The only thing I'll add to the DQ6 discussion is this: Beware the "Gigabyte Sata2 Controller" driver in Vista 64 :smile: The one on the website (from 04/07) installs just fine, but when you reboot, (even in safe mode) Vista BSODs trying to load it (using the last restore point to get rid of the driver is required.) When I reinstalled I used the driver from the motherboard's CD and it seems to be fine. I've seen one other post at Hard Forum citing a similar issue with it. (It's really a JMicro controller/driver).

I've had odd issues with the one data drive becoming read-only on random partitions, randomly at boot, but on the same controller, haven't seen this problem on the hotplug eSATA drive (identical Samsung Spinpoint T 500GB.) I think the problem has more to do with Vista making 0,0 of that drive a boot drive than the controller, but it COULD be a driver issue.


As for Vista Sleep, I'm very suspicious that it was responsible for the loss of the array. It could easily have been the power cable issue too, unstable power is never good, but I'm not sure it was the +12 that was having trouble, it may have been just the power-on that was shifting from position.

But Sleep resulted in that sluggish performance, constant HD grinding, and I was getting odd freezes and BSODS randomly, but it DID seem to be after waking up from sleep, I'm not sure I ever saw it before sleeping on a clean boot. The hangs ranged from "the CPU meter gadget is frozen in place, and the system is frozen in place, but video and sound still are running, the cursor moves, but I can't interact with anything" to a full BSOD (that unfortunately flashes by quickly and I couldn't capture where...one time it was iaStor I think though (the intel driver) and is probably what made the Raptor seem to fail (a real failure shouldn't have BSOD'd, it should have degraded the array and kept running.)

Granted, the BSOD happened again and failed out during the rebuild after a clean reboot, and took out the other drive, but it's highly likely that the first BSOD (if the problem was with the RAID driver) corrupted data on the other disk too, and the rebuild became very problematic.)

As I said, it COULD be a real bad sector that I'm not hitting again (especially with the Raptor sizes cut in half since I put it in RAID10 with 2x 80GB drives), but I can't see an HDD bad sector BSODing the driver since that's what the driver should be looking for, and Intel's drivers are usually pretty stable (even if their chips aren't) :smile:

I for one will NOT be trusting Sleep to my system ever again, not matter what MS thinks they've patched. Odd behavior is worth struggling through to see if it works, array loss is not :smile: (Besides, while Sleep is nifty for laptops and Joe Consumer's eMachine, I'm not sure I'd ever trust ANY RAID array to any power-saving suspend mode short of simply spinning down the drives if idle...not even the $800 3Ware and Adaptec hardware RAID cards) :smile: Soft-off just seems like a bad idea for the nature of that type of technology :smile: You end up bringing the array up from cold start without first POSTing the controller. What happens if a head crashed during shutdown or spinup, or a motor failed when reactivating? You're trusting it all to the driver (and RAM that could be faulty) and cutting the firmware out of the mix.