On Wed, Feb 13, 2013 at 2:58 AM, Dave Cundiff <syshackmin@xxxxxxxxx> wrote: > On Tue, Feb 12, 2013 at 6:25 AM, Arun Khan <knura9@xxxxxxxxx> wrote: >> I used 'watch cat /proc/mdstat' to watch the rebuild progress and the >> progress bar showed completion (100%) mark. >> When I broke out of this session and thereafter did 'cat /proc/mdstat' >> I noticed that not only was /dev/sdb1 not added but /dev/sdc1 was >> also not part of the array anymore. With two failed devices, >> /dev/md0 was still working mounted on /mnt/md0. >> > > have you tried adding the --force option to assemble? I would leave > out sdb since its an empty drive. > > If that brings it online you can try a read-only fsck with -n to check > the consistency of your data. Yes did use --force to assemble but no joy. Fortunately, I have been able to recover the data due to sheer luck! I figured I was already in a hole so there was no harm in reconnecting the 'failed' disk before RMA'ing it. The disk (/dev/sdb) was recognized by the BIOS, the OS (Debian) did not report and DRDY errors on the device. The Events count for raid partition (/dev/sdb1) on this device reported > zero. So now I had three devices with Events > 0 So far so good ... mdadm --assemble --force /dev/md0 /dev/sdb1 /dev/sdd1 /dev/sde1 did initialize /dev/md0 as active! mdadm --add --force /dev/md0 /dev/sdc1 added /dev/sdc1 into the array and I got a fully functional array. Then I 'failed/removed' /dev/sdb1 from the array (the original failed disk), /dev/md0 was still functional with 3 disks. I connected the new hard disk, partitioned /dev/sdb1 to match size and partition id (fd), mdadm --add /dev/md0 /dev/sdb1 gave a fully functional /dev/md0 It was a shot in the dark and it worked! Do not RMA a failing disk in a hurry, it might still save your day. Thanks to all for your help. -- Arun Khan -- To unsubscribe from this list: send the line "unsubscribe linux-raid" in the body of a message to majordomo@xxxxxxxxxxxxxxx More majordomo info at http://vger.kernel.org/majordomo-info.html
