This is not good news. I guess i will do the send and move to the newer
disk and get another new disk for that mirrored pool.
On 06/28/12 01:02 PM, Jim Klimov wrote:
2012-06-28 20:37, Jan Owoc wrote:
I wasn't able to find a solution, so I now have a non-mirrored root
pool. Does anyone know how to override this sector alignment issue?
Apparently, only by creating a new rpool on the 4KB drive
and zfs-sending data to it from the 512b drive. The idea
was discussed in detail a few times in the past few weeks
on this list and/or on zfs-discuss.
What is strange to me is the news that ZFS would refuse to
attach a 512b disk as a mirror to a 4KB TLVDEV (the rpool).
I did not test this myself, but it sounds like a bug - a
512b disk should certainly receive 4KB writes safely and
quickly (the opposite is not necessarily true) ;)
Did you certainly test that it doesn't work? If so - I think
some noise is due on the lists ;)
Also, for a bit more safety against data corruptions on a
single-disk rpool, you might want to use "copies=2" (or =3)
and rewrite the data to the pool (i.e. zfs-send|zfs-recv
within the pool) - in this case the userdata blocks would
be cloned like metadata is, and corruption (or mis-read)
of a single sector would not cause big problems.
Saved me on my home NAS quite a few times - nearly every
scrub finds a few CKSUM errors on the old root disk :)
HTH,
//Jim Klimov
_______________________________________________
OpenIndiana-discuss mailing list
[email protected]
http://openindiana.org/mailman/listinfo/openindiana-discuss
--
Dr. Daniel Kjar
Assistant Professor of Biology
Division of Mathematics and Natural Sciences
Elmira College
1 Park Place
Elmira, NY 14901
607-735-1826
http://faculty.elmira.edu/dkjar
"...humans send their young men to war; ants send their old ladies"
-E. O. Wilson
_______________________________________________
OpenIndiana-discuss mailing list
[email protected]
http://openindiana.org/mailman/listinfo/openindiana-discuss
