[nSLUG] Best way to upgrade mirrored disks to larger capacity?
D G Teed
donald.teed at gmail.com
Wed May 18 23:35:52 ADT 2011
On Wed, May 11, 2011 at 11:15 AM, Stephen Gregory <nslug at kernelpanic.ca>wrote:
> On Wed, May 11, 2011 at 09:16:27AM -0300, D G Teed wrote:
> > There are SATA connectors for only 2 drives on the mainboard
> > and they are not hot swappable.
> Are you sure? Any SATA I have used is hot swappable even if you have
> to pull the cable. You may need to force a rescan to find the new
> drives. I use rescan-scsi-bus.sh tool from Debian's scsitools package.
I should have clarified, they are not practically hot swappable in our 1U
as the box is a rebuild - was once SCSI with a backplane and is now
SATA with no backplane. If it is like most systems these days, it screams
shuts down if the case lid is removed.
I'm playing around with some new 1 TB SATA drives at home and found
hot swap was supported, but the device came back as /dev/sdc
rather than /dev/sdb - automatically handled by the kernel.
rescan-scsi-bus.sh didn't work for SATA, but didn't need the
scan to get the new device live.
> I don't think you will need a live cd to reinstall grub. You should be
> able to do it after you copy the data. You may need to chroot into
> the new array and install grub to hd0 and hd1 so you should hit your
> new drive.
I was thinking rescue cd to move / and other partitions with files
that don't copy well while the system is up.
> To save some headaches I would recommend that you modify the new
> /etc/fstab to use UUID= instead of /dev/mdXX as the md devices might
> Yes, Debian 6 has converted my fstab lines already, and likewise for
Partitioning has come up as an issue I was not aware of before. They
partitions to be on certain boundaries with the Western Digital "Advanced
and the move to 4KB blocksize. Western Digital's info on it for Linux is
I found a nice page telling us how to do the math in fdisk and illustrating
the performance difference in having the partitions sitting at the right
Note how when using multiple partitions, the next start block must be 8
than the previous end block. This becomes a gap of 64 with logical
including a gap of 64 prior to the first partition in the extended
Is there something like this happening with Seagate or others?
-------------- next part --------------
An HTML attachment was scrubbed...
More information about the nSLUG