Ethan wrote:
On Thu, Feb 18, 2010 at 13:22, Victor Latushkin
<victor.latush...@sun.com <mailto:victor.latush...@sun.com>> wrote:
Ethan wrote:
So, current plan:
- export the pool.
- format c9t1d0 to have one slice being the entire disk.
- import. should be degraded, missing c9t1d0p0.
- replace missing c9t1d0p0 with c9t1d0 (should this be c9t1d0s0?
my understanding is that zfs will treat the two about the same,
since it adds the partition table to raw devices if that's what
it's given and ends up using s0 anyway)
- wait for resilver.
- repeat with the other four disks.
Sound good?
Almost. You can run into issue with size - slice 0 on EFI-labeled
(whole-) disk may not be sufficient to replace disk in your raidz1.
regards,
victor
This should be okay, I think. The overhead from truecrypt was 262144
bytes, so I have that much to spare on the non-truecrypted disks. An EFI
GPT is 34 512-byte LBAs at each end, or 34816 bytes total. So there
should be plenty of room.
By default ZFS creates s0 on EFI-labeled disk at offset of 256 sectors from the
beginning of disk. Also there's 8MB reserved partition number 8.
_______________________________________________
zfs-discuss mailing list
zfs-discuss@opensolaris.org
http://mail.opensolaris.org/mailman/listinfo/zfs-discuss
