I have been wondering about that for a few days now. I also have a 3 disk
raid1 array, and currently, backups have to be done statically, ie: all
files have to be closed. Currently, this require some daemons to be
closed during the length of the backup. By stopping the daemon, then
failing a disk, then restarting the daemon, I could cut down on down time
by lots.
I'm very interested in testing this new feature.
Nick
> i see. Yep, you can sortof do this, via a simple trick. You cannot remove
> a 'working' disk currently, but your above description justifies it ...
> the trick is: make the disk fail. The simplest way is to use somthing like
> this:
>
> echo "scsi remove-single-device 0 0 3 0" >/proc/scsi/scsi
>
> (the numbers identify the SCSI disk). I will add explicit support (with
> apropriate safeguards) for working-disk removal. It's really simple, it's
> just not there yet because there was no reason.
>
