PRO6 "Remove inactive volumes" after interrupted balance

Hans,

Nothing special, just commands basically the same since the RAIDiator days.  Note that some of the below commands assume the "techsupport" boot mode which is just BusyBox over telnet with a line to Netgear if they want to jump in.  User: root, pw: infr8ntdebug.  If you can work over ssh on a running system, I highly recommend that as it will be faster and you'll have all the updated tools and filesystem choices in a "full" Linux system available to you.

I'm pretty sure I stole these or similar from the forum, Stack Overflow or Super User.  I'm absolutely no expert...

*** Only probably necessary in techsupport recovery, but see if you have
*** /dev/md/, /dev/md0, and /dev/md127 already detected & assembled
*** ready to mount - I only used --assemble --scan which was likely redundant
echo DEVICE partitions > /etc/mdadm/mdadm.conf
mdadm --examine --scan >> /etc/mdadm/mdadm.conf
mdadm --assemble --scan
*** End recovery commands
mkdir /mnt/data
mount -t btrfs -o ro /dev/md127 /mnt/data
*** mkdir /mnt/sys
*** mount -t btrfs /dev/md0 /mnt/sys

So really not much to it - just the "-o ro" for readonly allowed me to mount and copy my stuff to some Windows shares I mounted using CIFS.  md0 is the 4GB system partition, md127 is data, and md1 is swap I believe.  Just leave md0 or your system partition unmounted for now unless you can't boot to ssh...  And if you can boot to ReadyNAS OS there's no need to mount your system!

What I learned later were the commands above to be able to mount data without resuming the balance, the somewhat risky nature of running btrfs check (on unmounted data volume) to repair the filesystem, and the "btrfs balance cancel /mnt/data" which let me reboot normally without resuming the balance operation.  While I had the volume mounted in r/w I deleted a few things that I backed up just in case the balance decided to kick in again after reboot...

Please research any commands before you run them to make sure they're applicable and know what they do in your case.  Like, you usually won't need to configure mdadm.conf or assemble your RAID in normal mode!

So, if I were doing this over again, I would:

Boot up normally to an unmounted data volume but running ReadyNAS OS.

See if I can mount data volume read-only and back up everything necessary, incl ".apps" and related folders.  This may involve just copying everything "by hand" over the command line or using Midnight Commander.

If volume won't mount, check btrfs problems FAQ for other recovery mounts and tools. 

Run the btrfs check command without fixing to see what errors exist from my hard shutdown of the NAS :-(

If errors are minor, run btrfs check --repair to fix data volume (1 hour and up for each run for me)

Mount with -o skip_balance to avoid immediate balance resume.

Run "btrfs balance cancel [path to mounted volume]" to stop balance from restarting on next mount.

Lessons learned:

Either free up some space on the volume prior to the scheduled balance or run the balance manually.

Set up regular backup for the NAS for disasters and "volume abuse"

Techsupport boot mode is rarely necessary to use and most everything can be fixed in a running system...

Find out what forums the btrfs experts hang out in case I need them :)

I hope all this helps - it's more about the concepts than the copypaste here because one could potentially make things worse with a btrfs check.  That's why the readonly mount is a good thing if you need to make a current backup before the repair attempts.  I always go with the assumption that it's a goner and everything needs to be duplicated or restored before I try to fix anything.  Even if it turns out to be a minor repair, you have no way of knowing how serious it is before you start.