block rsv returned -28

- RAID10 btrfs volume consisting of 4 disks.
- One disk failed, was replaced, resync started
	(`btrfs dev add /dev/sdf /srv; btrfs dev del missing /srv`)
- Another disk failed before resync was done.
  Disk was replaced, resync restarted.
        (`btrfs dev add /dev/sdc /srv; btrfs dev del missing /srv`)

Naturally I don't expect it to recover from 2 failures, but
doing the attempt was worth a try, for this throws kernel warnings 
(below).

Can I somehow just have all inodes unlinked from the filesystem that are 
not recoverable? That would be preferable to re-mkfs.

The current status per fi show:
Label: 'srv'  uuid: 88300cd5-dbcb-4147-9ee4-c65a1c895e1d
        Total devices 6 FS bytes used 1.71TiB
        devid    7 size 1.36TiB used 29.50GiB path /dev/sdc
        devid    2 size 1.36TiB used 1.13TiB path /dev/sdd
        devid    3 size 1.36TiB used 1.13TiB path /dev/sde
        devid    6 size 1.36TiB used 377.00GiB path /dev/sdf
        *** Some devices missing

[ 1013.381148] btrfs: found 5451 extents
[ 1105.177601] btrfs: block rsv returned -28
[ 1105.177608] ------------[ cut here ]------------
[ 1105.177684] WARNING: at
/home/abuild/rpmbuild/BUILD/kernel-default-3.4.4~jng6/linux-3.4/fs/btrfs/extent-tree.c:6219
use_block_rsv+0xb8/0x104 [btrfs]()