Mirroring an entire volume group

[S.K. Chan]

Hi again David,
If you look at it closely it makes perfect sense because c2t1d0 has 2169 PEs and the corresponding c1t1d0 has 2170 PEs (ie one more than c2t1d0) so the arrangement or distribution of the extents are expected.
..
a) c2t1d0-2168 c1t1d0-2168
b) c2t1d0-0000 c1t1d0-2169
c) c2t2d0-0001 c1t2d0-0000
....
At a), c2t1d0 (total 2169 extents) reaches its last extent, therefore it continues on to the next PV (ie c2t1d0 starting at index 0000). However the corresponding mirrored extent on c1t1d0 ends 1 extent further that c2t1d0, thus it'll only continue at c) with c1t2d0 after the last extent is used in b).

This is perfectly what would be expected if you got PVs pair which do not have the identical number of extents. Bottomline everything looks good so far (from what I see).

[David Owens_1]

Thanks a lot guys. Better safe than sorry. Hopefully, I can end this lengthy thread.

David

[James R. Ferguson]

Hi David:

One comment. *Every* disk in an LVM volume group has a "LVM Header". If not, why would you ever need to do a 'vgcfgrestore' to a physical disk that you just replaced? :-))

One indication of this fact is the 'VGDA' count is a 'vgdisplay' matches the number of physical disk in the volume group.

Regards!

...JRF...

[David Owens_1]

Thanks James. I was just a little confused because I assumed that since all 8 disks were the same size and part number, they should have the same number of physical extents. And to live in grace and harmony, should the match? I belive that has been answered.


Thanks,
David