On Oct 29, 2013, at 8:37 PM, Edgar Fuß <e...@math.uni-bonn.de> wrote:
> I just tried to snapshot-fsck a 6T WAPBL filesystem (fsck -n -f -T ffs:-X > ...). > For about one hour, the RAID the filesystem is on was nearly fully busy. > The fsck_ffs process was consuming (just a little) CPU time, waiting on > biowait or tstile, but non-kill-able. > A ffsconfig -l was hanging in tstile. > After about one hour, the system panic'ed with "wapbl_register_deallocation: > out of resources". Replaying that filesystem's log took 40 minutes. > What can I do? Running that fsck command on much smaller filesystems works > fine. File system internal snapshots are expensive on large file systems as we have to build and clean a copy of all meta data. You could try file system external snapshots (-x /path/to/backup). -- J. Hannken-Illjes - hann...@eis.cs.tu-bs.de - TU Braunschweig (Germany)