I'm trying to understand exactly where on disk gmirror is going to write
its metadata, and was hoping I could elicit comments on whether I have
this right or not.
Assume my disk, da1, has a single slice, and I partition it as follows
(i'm making up sector sizes for the sake of simplicity):
a: 100 16 4.2BSD
b: 10 116 swap
c: 626 0
d: 500 126 4.2BSD
If I then use 'gmirror label -v -b round-robin gm0 /dev/da1s1', my
belief is that gmirror will write its metadata into sector 499 (the
last sector in the slice, which falls in partition 'd'). Thus, the
disk I want to mirror (da0) should have a 'd' partition that is 499
sectors in size. If it were 500, I would be asking for trouble down
Taking a slightly different approach, consider the recommendations here:
where the command to establish the metadata is
gmirror label -v -b round-robin gm0 /dev/da0
I believe that means the metadata will be written in the last sector of
the disk. Presumably this is part of the space left when fdisk is used
to set up the FreeBSD slice (assuming a single slice - and i'm assuming
there is virtually always space left), so the recommendation (setting up
gmirror on your live system disk) should work. Have I got that right?
Assuming a single slice configuration, is there a reason to prefer
one approach over the other? (i.e. metadata at end of slice versus
at end of disk)?
thanks in advance,
firstname.lastname@example.org mailing list
To unsubscribe, send any mail to "[EMAIL PROTECTED]"