Dell RAID 5 parity type?

[quasimodo]

Hello,

I am having some problems with a RAID set from a Dell server. I belive it's either a PERC4 or PERC5 controller. The firms IT guy took one of the 3 drives from the RAID5, then formated it and installed Windows OS on it.

I have tried de-stripping the RAID with various software tools, but nothing can read the strip set and the parity.

After using a formula to calculate the stripe set I concluded it's 128 sectors. But I still have two problems.

1. I don't know the exact order of the drives. I tried looking at the RAID header, but could not find a numeric marking for each drive. Either it's not there, or I was looking at the wrong place.

2. I can't find the parity type. After trying all software tools and looking at too much hex, I am somewhat lost. I did some research and found out that the PERC supposetley uses a proprietary algorithm.

Any help is appreciated.

Thanks!

[mediaman]

i'm sorry but if the IT guy took JUST one drive out, just buy another the same and put it in, the controller will do the rest. Raid5's can live without 1 drive.

Did you explain wrong?

[quasimodo]

I, know, but I don't have the controller.

[quasimodo]

I just have the drives, and even if I buy the controller, there could be some issues mounting and re-building the RAID due to different Firmware Rev. on the RAID controller.

[mediaman]

what software are you using to "virtually" create the RAID unit? There are very few standard stripe sizes, and order should not matter. Get the formatted windows disk off your IT guy and recover that too. Don't let him keep writing to it. (get it quick)

[quasimodo]

Well, as I said I determined the stripe size it's 128 sectors.
I tried several tools including R-Studio, WinHex etc... From my experience if your tools can find the parity and stripe set then it will usually also arange the drive order. But if this information can not be detected, then you have to supply it and or manually de-strip the RAID Array. I tried the common parity algorithms and no data was the result. As I have said I belive that Dell PARC uses a proprietary algorithm for it's parity. Therefore the tools are no good for this (automated) recovery.

Someone please correct me if I am wrong.


P.S I have images of all drives (including the one where the IT guy got really smart on)

[quasimodo]

Ok, I was able to get the original RAID card. It was actually a CERC card instead of PERC as I had asumed. I was able to recover the Data from the RAID set, but it still left me with some questions.

I had the right Drive order, but still had some problems finding the parity for the CERC cards. For some reason the 4 common parity types did not read the RAID.

Any information available on parity for CERC and PERC (aka adaptec) cards?

[drc]

Have you seen Runtime's Raid Reconstructor? I've had success with it.

[quasimodo]

yep, tried it, but It could not figure out the parity and therefore was not able to recover the data with it.

[drc]

now that you mention it I guess it does usually want to see all of the drives.

[N.C.]

Ok, I was able to get the original RAID card. It was actually a CERC card instead of PERC as I had asumed. I was able to recover the Data from the RAID set, but it still left me with some questions.

I had the right Drive order, but still had some problems finding the parity for the CERC cards. For some reason the 4 common parity types did not read the RAID.

Any information available on parity for CERC and PERC (aka adaptec) cards?

The parity structure for raid sets is really symple.
i think nobody needs to build own structure after 4 standard is already there....
I can only imagine the byte order for difference.

Anyway, if i need to simulate one raid set, i always use linux kernel, for this purpose.
This is the best solution, i think.

You can build on the fly all standard types, with all 4 parity types (left-asymmetric, left-symmetric, right-asymmetric, right-symmetric), without writing anythig to the disks.
you can image the constructed raid set simulations, and so on...
If the drives have headers, you can cut it (align the data area) with loop, and if the FS is linux based, the cowloop is the best solution for trying to repair the broken fs on huge arrays without making ANY modification on the sources.

If you interested, just pm me.

Regards,
Janos

PS:
Some day before i have read this on some guy's signature:
-------------------------------------
Microsoft is not the answer!
Microsoft is the question!
And "no" is the answer!
-------------------------------------

[apprent]

NC, I also prefer Linux for RAID troubleshooting.

But I can't seem to find a Linux distribution / LiveCD which would offer recent kernel + software and cowloop (they all offer device mapper, which is OK, but doesn't have some of cowloop's features...). Not even the LiveCDs which focus on system recovery (e.g. sysresccd).

[N.C.]

NC, I also prefer Linux for RAID troubleshooting.

But I can't seem to find a Linux distribution / LiveCD which would offer recent kernel + software and cowloop (they all offer device mapper, which is OK, but doesn't have some of cowloop's features...). Not even the LiveCDs which focus on system recovery (e.g. sysresccd).

I have not talked about live cd.
I usually use (installed) linux for recovery on my recovery-dedicated pc.

Regards,
Janos