Raid array being seen as 2 individual drives

Question

Hi. Here is the issue as posted in other places. Still searching for the answer to this one.
Specs:
K7n2 delta2 platinum with b50 bios
2x1gb crucial pc3200 2.5cas ram
AMD Barton 2500
2 x 160gb 7200rpm 8mb cache SATA Samsung Hdd's
Thermaltake 430w psu
Gainward fx5700 ultra graphics
OS's: original xp corp, slipstreamed xp corp sp2
raid drivers: nvraid.sys v4.27, 5.10, 5.11 (also the needed nvatabus.sys with those)
I am NOT overclocked.
fsb 166
1:1 ram/cpu
no spread spectrum or other garbage
ddr400 patch disabled
PSU gives presumably stable reading (according to what I see), with amperage ratings above the required.
checked and rechecked cables for bad ones
ran mulitple scans on drives, all come up drives OK
I HAVE installed into Raid 0 already, this is not an issue of hardware failure as far as I am concerned.
So here is the scenario
I have properly set up the array, using correct bios settings and the raid setup utility, for a raid 0 array of those 2 hdd's listed. When booting into xp, either version, I have used all 3 of the driver sets listed. I have been reinstalling to do some performance tests on different configurations.
Anyway, for the last few nights I have been trying to get the windows setup to see the raid 0 array as one 300gb drive. It does not, no matter what I try. It sees them as 2 drives, each being 160gb (or thereabouts). These drives are matched, same firmware, same lot, so that should not be an issue.
I have used numberous tools to delete the mbr on the drives, both in an array and as single drives. I have done the same as well as tried an install and formatted each drive individually, still the same effect when the raid array is recreated.
Basically, I can find no good reason why the array is seen as individuals and not as an array. It is interesting to note, that even though xp setup sees the the array as 2 drives, I can complete the text based portion of setup. However, rebooting to start the GUI portion of setup, it will not boot. Obviously becuase the bios has the controller as the nvraid controller and it is supposed to be a raid 0 array, so I expected that.
Short of rewriting the mbr, either by deleting it or by changing each drive by formatting/partitioning/installing an OS on them, I cannot think of how to fix this. I know the drives and xp cd's work because I have already installed with them.
I understand what to do in the bios portion, and in the raid setup utility portion. I know that I can boot into windows as a single drive and use the nvraid tool to set it up, but that is not the way it should be, and that is not the way I am going to learn WHY this is happening.
Roger that. First set in bios enable raid (in this bios I have to enable IDE array, then choose which controller to actually enable raid on, which happens to have been SATA 1 & 2).
Second, upon reboot, I use the F10 key to enter raid utility. Then, set to striping, set stripe size (which was one of the things I am testing), and add the drives to the array. Next step is to create it. It asks to clear disc data, and it is done.
Have deleted that array as well as just cleared it. Have deleted it and reboot and rebuild it. Have deleted it, reboot, change bios back to non-raid, reboot. Reboot. Change bios back to raid enabled. Reboot. Rebuild array in raid utility, reboot. Run setup, only see 2 hdd's, not one array.
Umm, yep, that is about it.
More to the story now.
From some other posts I tried this.
1. destroy array. reboot. disable raid in bios. reboot. verify sata's visible as singles in bios.
2. power down. pull plugs on sata's. reboot. no drives visible.
3. pull power. jumper clear cmos. wait 60 seconds. re-pin jumper. power up.
4. verify no drives. verify default bios settings. all is good
5. plug drives in. reboot. seen as singles. erase mbr on both drives. reboot
6. enable raid in bios, and choose sata 1 & 2 as "enabled". reboot.
7. use F10 key to setup raid. Here is the interesting part. Even though I deleted the array prior to all of this, and removed the drives to force an ESCD update, and cleard the cmos with the board jumper, and then before raid was enabled, cleared the mbr on the drives, when I started the raid utility, the array was already set up. That is the problem, whatever that is. I have read snippets where it is claimed that this chip or bios or whatever stores some kind of a table on this stuff, but this is a bit out of hand.
That combination, IMO, should have cleared anything out. But, the saga continues.
Thanks for you help BWM
[Edit] BTW, I have finally found a utility that will see a raid array and allow me to clear the arrays mbr. It is called SuperFdisk and is at ptdd.com. So far the only one that see's the 2 drives as 1.
Yeppers.
Started with v5.10 which came on a floppy with the mobo. Told setup to use both, nvatabus.sys and nvraid.sys. Even switched which one of the 2 I picked first, just to see.
Same thing with v4.27 and v5.11. Also tried it with just the nvraid.sys and just the nvatabus.sys (which obviously does squat for raid, lol)
Trying some new things now. Post in a little bit.
I am officially at 'Wit's End'.
Here is what I have tried now.
1.pull drive cables. pull power. jumper clear cmos. wait. power up. no drives
2.plug sata 1 in. boot. drive detected.
3.boot to command.com, run MHDD, which is a nice russian utility similar to Spinrite. Used this to clear the mbr at hardware level, and do a complete erase.
4. reboot to command.com. run superfdisk. erase mbr.
5. pull plug on sata 1, and plug in sata 2 with sata 1 cable. repeat the erasure steps listed above.
6. pull plug on sata 2, no sata plugged in. reboot
7. change bios to raid enable on sata 1 & 2. power down
8. plug in sata 1 & 2. power up.
9. inspect raid utility. no listing of any arrays. reboot
10. in raid utility, build array. did NOT clear discs. reboot
11. attempt install. single drives found again (used both drivers).reboot
12. in raid utility, optioned to CLEAR discs (funny, rebuild option is never valid).reboot
13. attempt install, both drivers, still seen as 2 individuals.
Things to note. When creating an array when presumably there are none, it assigns the raid array an ID of 2. Upon reboot, the ID is now 1. Don't know what difference that makes.
Also, tried the install listed above with APIC functionality both off and on. Also, when on, set MPS to both 1.1 and 1.4. In addition to this, each variant I tried manual HAL layers of, in this order, ACPI (the one that actually spells ACPI out), ACPI Uniprocessor, MPS Uniprocessor, and let it choose it for me.
So, here I sit in a barca-lounger at 'Wit's End', with a warm cup of java and a dinner mint.

Answer 1

Here is the final product on the floppy disk that I used to  successfully install a stable raid 0 on the MSI K7N2 Delta 2 Ultra 400  Platinum ms-6570e motherboard.
On root of floppy, from driverset 6.70. (after much testing, I used  driver pack 5.10 for my nic and smbus. I used the realtek sound  drivers off the cd for audio. I have used every driver pack I could  find, and while some did offer better I/O or read/write latency, this  set in general provided the most stable environment. The only drivers  I used were these floppy drivers for SATA, the nic and smbus just  mentioned, the sound just mentioned, and updating the nvide drivers to  mside drivers)
<from sataraid directory>
disk1
idecoi.dll
nvatabus.sys
nvraid.cat
nvraid.inf
nvraid.sys
nvraidco.dll
<from legacy directory>
nvata.cat
nvatabus.inf
I used the txtsetup.oem from the sataraid directory, but edited this:
[Files.scsi.RAIDCLASS]
driver  = d1,nvraid.sys,RAIDCLASS
inf     = d1,nvraid.inf
dll     = d1,nvraidco.dll
catalog = d1,nvraid.cat
[Files.scsi.BUSDRV]
driver = d1,nvatabus.sys,BUSDRV
inf    = d1, nvraid.inf
dll    = d1,idecoi.dll
catalog = d1, nvraid.cat
To this:
[Files.scsi.RAIDCLASS]
driver  = d1,nvraid.sys,RAIDCLASS
inf     = d1,nvraid.inf
dll     = d1,nvraidco.dll
catalog = d1,nvata.cat
[Files.scsi.BUSDRV]
driver = d1,nvatabus.sys,BUSDRV
inf    = d1, nvatabus.inf
dll    = d1,idecoi.dll
catalog = d1, nvata.cat
Now, it is important to note that I installed or attempted to install  at least 50 times. Bare minimum. I noticed when I use this custom  driver disc that in the GUI portion of setup, XP asks me for files  from the disc. I tried lot's of different things to alleviate this,  and denied some of them.
One thing that really bugged me was that the bios would see my #2  optical, slave on secondary IDE channel. A dvd/rw drive. And I could  even start the setup from it. But, once I got about 3/4 through copy  file stage on text setup portion, I would hang. Becuase the drive was  no longer accessible. Booting from the master would get me to the  desktop, but the slave optical was nowhere to be found. Updating the  ATA/IDE controller to the ms ide drivers would get it visible, but I  kept having issues with stability after I did that.
The most stable method I found was to use my above listing of driver  files for the floppy, and when in GUI mode setup asks about NVCOI.DLL,  I skipped it, ignored it, and did not let setup install it. That  actually got me to the desktop, with access to the slave optical as a  "removable drive". It even knew what the hardware was. It just could  not access it. On a reboot however, back to not seeing it. This method  however did allow me to update the nvide driver with the mside driver  with no stability issues. So, for me it was a raving success.
Here are some links regarding the SATA RAID driver workaround:
http://66.102.7.104/search?q=cache:jHbX5bNfGx4J:www.msfn.org/board/lofiversion/index.php/t51140.html+nforce2+nvraid.sys+ms+ide&hl=en&client=opera
http://www.aoaforums.com/frontpage/index.php?option=content&task=view&id=292&Itemid
http://66.102.7.104/search?q=cache:J9UhG2Kd8W4J:www.short-media.com/forum/showthread.php%3Ft%3D32751+xp+2+sata+raid+0+seen+as+individual&hl=en&client=opera
Early on one problem I noticed was that in text setup mode of xp  installation, there were long pauses that I have never seen before. I  noticed that with both ide and sata installs. Also I noticed that when  booting there was a really long pause when the xp logo is first seen  in a sort of dim state till when it became bright and vivid.
Come to find out that this is a more or less typical scenario. Most  instances that I read about were all pointing to the nvide driver. So,  I found if I just updated the PATA controller to the standard ms ide  driver, that went away and the whole system ran better.
It took awhile to figure out that if you install a driver with the  nForce2 chip, you had to uninstall it or you will have issues. Herein  was the main problem I encountered with the SATA RAID installs. The  nvatabus.sys driver was required for an SATA RAID install. Omitting  the ata driver was impossible. And for awhile I had no success  updating the ms ide driver once I was to the desktop without major  instability. Here are some links regarding the drivers for this  chipset:
http://www.nforcershq.com/forum/latest-drivers-for-nforce-3-vt60240.html
In my browsing I came across some pretty interesting articles  regarding ACPI. One thing I started playing with was the different HAL  layers that xp installs on it's own, vs. me picking one manually (F5  key). I must have started the setup at least 50 times to figure out  this: that this particular board does not give me the bios settings to  install xp with anything but the ACPI Uniprocessor Hal. For instance,  the MPS Uniprocessor HAL is much more responsive, but it lacks the  IRQ's needed for setup to see the raid array. I booted to each one,  some locking the system up, some booting OK. The one I found the best  performance with early on was the one that spells out ACPI, not just  initialized. (sorry, I don't want to look it up).
I seemed to be getting closer, but I could not find the needed bios  settings to properly manage my ACPI, and since I was trying for RAID,  I could not use the one that did work. Here is a link for that kind of  stuff.
http://www.fceduc.umu.se/~jesruv98/info/acpi/acpi.html
Another thing that I did not like was being forced to use the dynamic  overclocking feature of this board. I have a 333mhz barton core, and I  have ddr400 ram. In optimized (fool proof) mode in bios, I was running  asynchronous. I did not want that. So I set it down to run at 166mhz,  with very slow and conservative settings on everything. Unfortunately,  if I did this "manual" method, I was forced to use the dynamic  overclocking. I thought I had that figured out. So I set everything to  "optimized". But, as it turns out, the system had terrible stability  without the dynamic overclocking set to at least Private. What this  meant is that I could not rule out that my stability issues  (corruptions and hangs and bsod) were from being overclocked even a  tiny bit or not. And as if that were not enough, this bios has a  special set of settings you must unlock to see. And one of those is  paramount in achieving a stable system. It is called the DDR400 patch,  and it is enabled by default. So, by pressing SHIFT+f2 AND CTRL+F3,  these settings are now available. Like I said, I had to disable that  DDR400 patch setting.
I also found out from the first day that my board shipped with the  latest bios. I flashed the 2 prior versions with no success in more  stability. After about 6 weeks of getting whipped on by this board, I  found mention of some modded bios's for this board. I have used modded  bios's in the past, some worked wonders, others required some serious  effort to recover from. What I found out about this board is that  there are 2 players who make the modded bios's. Here is the first  index I found from a german website. This one actually is for the  older B4 version only for the Platinum.
http://storage-raid-forum.de/viewtopic.php?t=2824
And here is an english forum for pretty much the same thing
http://www.nforcershq.com/forum/bios-mods-for-k7n-and-k8-boards-vt55014.html
These links have a bit more information, and I decided to go with  these. I tried versions b61,b62 and b71. I found b71 to work the best  for me. Mind you I am not into overclocking or what-have-you. Just a  rig that performs as well as it was advertised to do. Try these out  for the bios information:
http://forums.pcper.com/showthread.php?t=385480
https://forum-en.msi.com/index.php?topic=84715.0B62
Here is a page that had a bunch of misc stuff I found interesting:
http://66.102.7.104/search?q=cache:QkvLeKcbwjQJ:www.amdzone.com/modules.php%3Fop%3Dmodload%26name%3DPNphpBB2%26file%3Dviewtopic%26p%3D75383+nforce2+ultra+nvraid+driver+freeze&hl=en&client=opera
In the end, I have, I think, conquered this board. My findings can be  summed up as follows, all in my opion only I guess.
1. There are some ACPI/APIC issues with this board or this chipset. I  believe it also included drivers and some can be attributed to XP.
2. There are some major bios issues with this board.
3. There are some major driver issues concerning SATA/RAID. I am not  sure who get's the boob prize, nVidia or MSI.
The only way I have found to get RAID 0 installed and stable is to  modify my bios (which is a modded beta version), modify my driver disk  for SATA/RAID, modify my install sequence for those drivers, modify my  drivers within windows after setup, use different drivers from  different driver packs for different pieces of hardware, and modify my  HAL layer after everything else is done, to achieve peak performance.
If I had not spent soooo much time trying to get a stable install, I  would have built up an Unattended CD, which has some possibilities for  forcing non WHQL drivers. But, hey man, I am totally burn out on this  board. And all it was for is a spare LAN box for when I go to a  lanparty. Sheesh. Murphy's law.
Oh, and I also found out, with my own eyes, that the Soyo KT600  Dragon+ that I dumped for this wonderful board, is way faster. Faster  read/writes, faster throughput on the nic, faster booting, much faster  installs of xp. As a matter of fact, I could get my KT600 to get a  consistent thruput on the network to my older KT266a board at 99%.  That is pretty fast. 2 of these Platinum boards, on a sweet switch  that is tweaked, will only go up to 91%, no matter how much I tweak  them. The gigabit connects via a crossover cable at about 38% of full  bore. This is tweaked stuff, but still. I listened to the hype. Dual  channel memory, giglan, etc etc.
I hope this may help anyone else out there who is still fighting with  these issues.
Out.
sul