Fusion DX800RAID Recovery

[kpeddie]

Hi

Anybody happen to know the standard RAID configuration on this box. 8 Disk RAID 5 array where disk one has failed(bad sectors and iffy head). While customer was rebuilding array another drive failed with what appears to be just bad sectors. Trying to rebuild from 7 disks but not had much luck yet. Block size, parity location etc would be handy if anyone knows

[lcoughey]

PM Sent

[dmarques]

Hi,

I never had one of those, but you can try to start checking that manually.
Find the type of file system first of all.

[kpeddie]

Its in HFS+, I managed to get the directory structure up in Rstudio but files are corrupt.

[dmarques]

So probably something is wrong with drive order or stripe size.

[ibcho]

i think they are sata hds, if they are, dont keep using original hdds , make clone of all hdds into good working ones, and try work clone ones

[hddguy]

You can easily find offset from start of data, also can usually tell first and last disk in RAID5, from here you can identify first and last disks physically and maybe they take some native order.

HFS+ have no sequential file counters, so if you dont know config you need to work it out from checking consistency in winhex, or through analysis of XOR blocks and pattern it takes.

[kpeddie]

This one is really beginning to annoy me. Drives came into us in the box and customer had numbered them. Its 8 drives in RAID 5 HFS+. Partitions show up on what is apparently drive number 8, so im a bit suspicious at that. Anyway, this box has an Atto R380 in it, im using stripe size of 128KB and left sync block order. When i put that into R studio with the drives ordered 8,7,6,5,4, etc then bang i get a file system showing up under Raid block, a few files are viewable but most are not. I suspect the drive order is wrong but I really have no idea how to work out drive order in HFS+. Any clues?

Reclaime Raid would seem to want to take about a year to run, left it for a couple of days and it had hardly made any progress.

Customer is not that willing to pay a huge sum for data so i may at this point just give up unless anyone has any bright ideas.

[kpeddie]

I should also add that I have tried order 1,2,3,4,5 and every stripe size between 1kb and 1mb just in case my calcs were wrong.

[kpeddie]

Well drive set fails XOR test in most places so something funny has happened... does pass in certain areas so im guessing a failed rebuild or customer has moved drives about and tried rebuild and quit partially the way though?

ah well

[dmarques]

Or else some drive was not part of the RAID already.
Have you tried to remove each drive from the RAID?

[dobrevjetser]

Customers often "forget" to say what they did with raid before coming to you.
Agree with dmarques. Try to leave a drive out.

Dobre

[kpeddie]

Hi

Already tried that but made no difference, well, the hex within some of my test files changed but not in any helpful way....

[DR-Kiev]

Well drive set fails XOR test in most places so something funny has happened... does pass in certain areas so im guessing a failed rebuild or customer has moved drives about and tried rebuild and quit partially the way though?

ah well

Xor test could fail if one or few drives weren't woking periodicaly.
I would suggest you to reseach Meta data at the end of each drives and revers details about (offset-order-parityType-stripe_sizes).
Or use external help. For example ME