Well, unfortunately I found out that we make the disk sets by rsync, not by a disk image byte copy, so we don't know what all of the original bytes of the disks were. We could probably say what the starting offset of the partition was, but if the RAID label is any larger than a partition table, then the filesystem might have some corruption or overwritten data.
I'm not sure what your best option is at this point, but the connectome in a box page says that we provide checksums to check your data against (provided that you managed to restore the partition table, and it recognizes the filesystem), and you could download any missing or corrupt files via the REST interface or such. We might be able to work something out with sending a fresh set of disks and you returning those, for shipping costs or something, but I'm not sure who would make that call. Tim On Wed, Dec 16, 2015 at 12:41 PM, Timothy Coalson <[email protected]> wrote: > Creating any kind of raid array on a disk nearly always involves labeling > the disk with the raid configuration, which probably overwrote the > partition table (and is hidden from the exposed raid-0 "disk", causing the > "disk" to shrink and start at a different offset). Hardware raid > controllers are unfriendly to disks that already have data on them, we > should probably mention this in the instructions somewhere. > > We can probably make the first several MB of each disk available for > download somewhere, which may be able to repair the part of the disk the > RAID controller likely overwrote. You will need to attach them to a normal > controller (AHCI, or even an external usb 3 case) in order to attempt this, > or to use them at all. > > Tim > > > On Wed, Dec 16, 2015 at 10:42 AM, Bogdan Petre <[email protected]> > wrote: > >> Hi everyone, >> >> We've recently ordered connectome in a box and have been having some >> trouble accessing the data on the disks. The disks did not appear to >> have valid partition tables on them which could be recognized by our >> linux servers, and I was wondering if I could get some details from >> anybody who might know more regarding the partition scheme (GPT?), and >> how data is configured across the 5 disks. >> >> We've installed them in our RAID array and tried to bring them online as >> individual disks using a trivial raid 0 configuration (since raid 0 on a >> single disk should just be the same as a non-raid configuration). The >> disks became visible to our operating system but no partitions were >> detected (i.e. we can see /dev/sdc but not /dev/sdc1 for instance). >> Various disk partition management tools (fdisk, gdisk, parted, gparted) >> all complain about corrupted or invalid partition tables, and this has >> all been the case for both of the two disks I've tried of the five. >> >> If anybody else has had similar experiences or has any useful >> information to share I'd appreciate it. All I've found on the connectome >> website are instructions to hand this to our IT person (me, in this >> case) and that the drives are formatted with EXT3, neither of which have >> helped. >> >> Thanks, >> >> -- >> Bogdan Petre >> Department of Physiology >> Northwestern University >> >> _______________________________________________ >> HCP-Users mailing list >> [email protected] >> http://lists.humanconnectome.org/mailman/listinfo/hcp-users >> > > _______________________________________________ HCP-Users mailing list [email protected] http://lists.humanconnectome.org/mailman/listinfo/hcp-users
