Date: Fri, 22 Oct 2004 09:00:46 -0500 From: Joe Koberg <joe@osoft.us> To: secmgr <security@jim-liesl.org> Cc: freebsd-stable@freebsd.org Subject: Re: newfs -v on vinum raid5 panics system (4.10-p3 RELEASE) Message-ID: <4179128E.3050704@osoft.us> In-Reply-To: <41783571.5070204@jim-liesl.org> References: <41783571.5070204@jim-liesl.org>
next in thread | previous in thread | raw e-mail | index | archive | help
secmgr wrote:
> partition the same size as the slice (c=h), and then let the subdisks
> use the entire partition (len 0), the raid set is corrupted every time
> after initializing. This usually leads to a kernel panic during
> newfs. If I leave some amount free, (ie the subdisk only uses 8000mb
> of a 8675mb drive) no problem. If i'm RTFM'ing correctly, it looks
> like it should reserve 132kb per drive. I guess I assumed that would
> be automagically reserved, or if it's supposed to be, it's getting
> walked on.
>
> Next, I disklabel each drive (they're all identical)
> # /dev/da0s1c:
> # size offset fstype [fsize bsize bps/cpg]
> c: 17767827 0 unused 0 0 # (Cyl. 0 -
> 1105*)
> h: 17767827 0 vinum # (Cyl. 0 -
> 1105*)
>
I alway leave 64 sectors free at the beginning of the disk:
h: 17767763 64 vinum # (Cyl. 0 -
1105*)
I think this is where the partition tables live, you are probably
overwriting them when vinum writes its metadata. I think a normal
UFS filesystem will leave 64 sectors of slack space at the beginning
of the partition, but vinum may not. I may also be totally wrong.
Joe Koberg
joe at osoft dot us
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?4179128E.3050704>