Re: Recovering partitions from disk image? *resolved*
so write a short article about how you did this and why using hardware RAID solutions is bad, and put it on your website. it's AT LEAST funny that your hardware raid instead of protecting - rendered your data inaccessible. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: Recovering partitions from disk image? *resolved*
If anyone's interested, the last post I made about doing a bsdlabel, fsck, mdconfig etc on the damaged disk image worked. I have recovered all my files! Hooray!!! Skye -- View this message in context: http://www.nabble.com/Recovering-partitions-from-disk-image--tp22862006p22889910.html Sent from the freebsd-questions mailing list archive at Nabble.com. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: Recovering partitions from disk image?
Hello FreeBSD gurus, I recently had the pleasure of trying to recover a failed RAID1 array. It consisted of two 120GB disks in mirrored configuration. Both drives have a ton of bad sectors, so bad that the 3ware RAID card stopped recognizing that there was a mirror at all. Having no other options that I could think of, I excellent hardware - designed to PROTECT ;) pulled the drive with the one with the least read errors, directly connected it to ATA bus and used GNU ddrescue to make a disk image with only about 7 read errors. Neither fdisk nor bsdlabel can read the disk image. I wonder if there's some funny data at the beginning of the hard drive the 3ware card used for RAID configuration? If this is true, is there a way to search for the beginning of the real fdisk data and lop it off to make a possibly valid disk image? do dd if=image bs=128k count=1|hexdump -C|less and look for it :) at MBR sector you'll see probably something at the end like this: 0190 46 0a d0 e3 00 5e 05 28 46 02 77 88 c3 52 65 61 |F.##.^.(F.w.#Rea| 01a0 64 00 42 6f 6f 74 00 20 65 72 72 6f 72 0d 0a 00 |d.Boot. error...| for bsdlabel it's easier it's WEV and first 3 bytes. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: Recovering partitions from disk image?
Update: I figured out how to get scan_ffs to read a file by looking at the program source (if it starts with / then it considers it a regular file to read instead of a device) and got the following results which matches well with the TestDisk output. $ scan_ffs -s /recovery/disk0.img ufs1 at 1087 size 2621440 mount / time Wed Dec 31 16:00:00 1969 ufs1 at 10486847 size 5242880 mount /var time Wed Dec 31 16:00:00 1969 ufs1 at 31458367 size 5242880 mount /usr/home time Wed Dec 31 16:00:00 1969 ufs1 at 54525634 size 46680873 mount /mnt time Wed Dec 31 16:00:00 1969 Looks about right compared to the df output I had from that host: Filesystem Size Used Avail Capacity Mounted on /dev/twed0s1a 4.9G 2.8G 1.7G62%/ /dev/twed0s1e 9.8G 5.0G 4.0G56%/var /dev/twed0s1f 9.8G 952M 8.1G10%/usr/home /dev/twed0s2e88G15G65G19%/mnt So, what can I do with those numbers? It doesn't look like there's any valid MBR or disklabel on this disk image. Can I extract these filesystems one at a time from the image and mount them somehow? Thanks, Skye -- View this message in context: http://www.nabble.com/Recovering-partitions-from-disk-image--tp22862006p22872988.html Sent from the freebsd-questions mailing list archive at Nabble.com. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: Recovering partitions from disk image?
On Fri, 3 Apr 2009, snott wrote: Update: I figured out how to get scan_ffs to read a file by looking at the program source (if it starts with / then it considers it a regular file to read instead of a device) and got the following results which matches well with the TestDisk output. $ scan_ffs -s /recovery/disk0.img ufs1 at 1087 size 2621440 mount / time Wed Dec 31 16:00:00 1969 ufs1 at 10486847 size 5242880 mount /var time Wed Dec 31 16:00:00 1969 ufs1 at 31458367 size 5242880 mount /usr/home time Wed Dec 31 16:00:00 1969 ufs1 at 54525634 size 46680873 mount /mnt time Wed Dec 31 16:00:00 1969 Looks about right compared to the df output I had from that host: Filesystem Size Used Avail Capacity Mounted on /dev/twed0s1a 4.9G 2.8G 1.7G62%/ /dev/twed0s1e 9.8G 5.0G 4.0G56%/var /dev/twed0s1f 9.8G 952M 8.1G10%/usr/home /dev/twed0s2e88G15G65G19%/mnt So, what can I do with those numbers? It doesn't look like there's any valid MBR or disklabel on this disk image. Can I extract these filesystems one at a time from the image and mount them somehow? Looks like scan_ffs is reporting block size. I'd take a spare computer with a blank disk and do a minimal FreeBSD install on it, setting the units to blocks in the partition screen and duplicating the values given by scan_ffs. Then connect your read-only image media and dd to each partition, using dd's skip option to skip over previous partition data. Daring users (i.e., those who are sure they won't confuse source with target) might install a blank drive in a live machine and use fdisk, bsdlabel, and dd from there. -Warren Block * Rapid City, South Dakota USA ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
Re: Recovering partitions from disk image?
Warren Block wrote: Looks like scan_ffs is reporting block size. I'd take a spare computer with a blank disk and do a minimal FreeBSD install on it, setting the units to blocks in the partition screen and duplicating the values given by scan_ffs. Then connect your read-only image media and dd to each partition, using dd's skip option to skip over previous partition data. OK, I can see how that would work. I don't really have a spare computer/drive to muck around with though, at least not easily. If I can work directly on the disk image that would be much better. Could I fix the bsdlabel on the disk.img and mount it with mdconfig? I noticed scan_ffs can output a bsdlabel config file. Something like: # scan_ffs -sl disk.img disk.label # bsdlabel -w -f disk.img auto # write a fresh default label # bsdlabel -e -f disk.img # and load in scan_ffs output # mdconfig -a -t vnode -f disk.img -u 0 If that looks like it might work, should I fsck the disk image before or after mounting with mdconfig or not at all? Do I risk kernel panic without fsck? Thanks, Skye -- View this message in context: http://www.nabble.com/Recovering-partitions-from-disk-image--tp22862006p22878294.html Sent from the freebsd-questions mailing list archive at Nabble.com. ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to freebsd-questions-unsubscr...@freebsd.org
