2011-12-17 21:59, Steve Gonczi wrote:
Coincidentally, I am pretty sure entry 0 of these meta dnode objects is
so the block with the checksum error does never comes into play.
I wonder if this is true indeed - seems so, because the pool
seems to work reardless of the seemingly deep metadata error.
Now, can someone else please confirm this guess? If I were
to just calculate the correct checksum and overwrite the
on-disk version of the block with "correc" one, would I
likely make matters worse or okay? ;)
Thanks to all that have already replied,
zfs-discuss mailing list