I just went through this twice twice in the last month. I kept notes but I dont guarentee they are 100% accurate!
In both cases what I did was extract the faulty drive(s), install replacement drive(s), reload SME, then attempt to salvage the data by attaching the (or one of a mirror)device to a spare IDE/SCSI cable connection. This worked 100% both times and, again, I tip my hat at SME and how simple solutions and (now) recovery is with the product.
Anyway, onto the specifics:
1) Attach the faulty device to the operational SME & reboot
2) Drop down to the root prompt (alt-F1/f2/f3) and sign in as root
3) Use fdisk /dev/hdb (IDE) or /dev/sda (SCSI) , substituting the correct identifer name for your situation. A cat of the /etc/fstab should reveal your existing drives (I can recall if the added one shows ) . The new (guest) drive is normally one alpha letter higher than the resident disk names (hda,hdb,hdc for IDE, sda,sdb, etc for SCSI) then use the "p" option to print the partition table. In my last case it looked like this:
Device Boot Start End Blocks Id System
/dev/sda1 * 1 2 16033+ 83 Linux
/dev/sda2 3 522 4176900 5 Extended
/dev/sda5 3 35 265041 82 Linux swap
/dev/sda6 36 522 3911796 83 Linux
The "dead" disk is the sda5/sda6 combination , the data clearly (by the size) is the sda6 partition.
4) I created a new folder to act as a mount point with a "mkdir oldsmeserver".
5) I then mapped the above data partition of the old drive to this new mount point with a "mount -t ext2 /dev/sda6 oldsmeserver".
6) The old disk now appears within this newly created folder and I was able to copy/untar etc the content within.
The other option I have used in the past is to hook the SME disk up to a Lindows/Linspire workstation (linux) and pull data off using a visual tool. That trick once worked for an SME software mirror set that was so corrupted neither would boot on the server nor would it even be accepted by the technique above.
Hope this helps!
3 Replies
1413 Views