Date: Mon, 27 Feb 2017 03:09:18 +0300 From: Andrey Fesenko <f0andrey@gmail.com> To: Zaphod Beeblebrox <zbeeble@gmail.com> Cc: FreeBSD Hackers <freebsd-hackers@freebsd.org> Subject: Re: ZFS should prefer the most specific... or something. Message-ID: <CA%2BK5SrP=p7xqFOjPM4BvDuyitM951biu8ZNoKuin8A_TfeG1xQ@mail.gmail.com> In-Reply-To: <CACpH0Meg5MjOGCohSMrAdwdRG_smVNY6CNVRaUSXw9cq0x2YsA@mail.gmail.com> References: <CACpH0Meg5MjOGCohSMrAdwdRG_smVNY6CNVRaUSXw9cq0x2YsA@mail.gmail.com>
next in thread | previous in thread | raw e-mail | index | archive | help
On Mon, Feb 27, 2017 at 2:31 AM, Zaphod Beeblebrox <zbeeble@gmail.com> wrote:
> So, my experience is that I added two drives (that showed up as ada0 and
> ada1) ... something odd with the raid card added a da0 ... which is
> unreadable ... seems to be a bogon.
>
> ... and ZFS, in it's infinite wisdom changes my nice, neat da0s2a zfs to
> diskid/<blahblahblah>s2a.
>
> ... this is doubly bad, because then the /dev/da0s1a cannot mount as swap.
>
> I realize that this is can be somewhat fixed for non-root zfs with
> import/export, but for root ZFS ... you're stuck.
>
> Can we have ZFS prefer the shortest (rather than longest)? Can we have it
> prefer labels, if they exist (have a priority field in the different types
> of geom ... an order in which they are tasted)?
>
> The current situation has been annoying multiple distinct times.
"A disk can be specified by a full path to the device or the geom(4)
provider name" see zpool(8)
For example gpt lables
gpart add -a 4k -t freebsd-zfs -l local0 da0
...
% zpool status localz1
NAME STATE READ WRITE CKSUM
localz1 ONLINE 0 0 0
raidz1-0 ONLINE 0 0 0
gpt/local1 ONLINE 0 0 0
gpt/local2 ONLINE 0 0 0
gpt/local0 ONLINE 0 0 0
logs
gpt/l2arc ONLINE 0 0 0
cache
gpt/cache ONLINE 0 0 0
Label or disk name may be changed if need after export/import (If I
remember it right)
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?CA%2BK5SrP=p7xqFOjPM4BvDuyitM951biu8ZNoKuin8A_TfeG1xQ>