On Fri, Nov 2, 2012 at 8:52 PM, Bob S <
robert...@alcatel-lucent.com> wrote:
> I had to add an lsmod test and a "modprobe fuse " in the startup script to
lsmod shouldn't be necessary. modprobe-ing a loaded module will return sucess.
Or, just put fuse in /etc/modules to get it always loaded.
> a writeup would be good.
The thing is I don't think anyone is maintaining zfs-fuse anymore.
> My concern is that if I drop in more drives, the /dev/sd* values could get
> swapped around. This would make things very ugly for zfs.
> On other scsi linux systems I use the LABEL parameters to mount drives
> and not worry about device mappings.
>
> I use ZFS on Solaris regularly, on that system the commands look like
> zpool create -f zfspool mirror c0t2d0 c0t10do
> zpool add -f zfspool mirror c0t4d0 c0t12d0
> In this case, unless I put the drive in the wrong slot, the mapping will
> always work.
>
> Does anyone have any experience with this sort of setup?
>
> Is there documentation which I have not stumbled across which describes how
> to
> handle this sort of situation?
From the other related project:
http://zfsonlinux.org/faq.html#WhatDevNamesShouldIUseWhenCreatingMyPool
The first three should apply to zfs-fuse as well.
--
Fajar