Skip site navigation (1)Skip section navigation (2)
Date:      Sun, 16 Sep 2001 02:08:28 +0200
From:      Bernd Walter <ticso@mail.cicely.de>
To:        john_wilson100@excite.com
Cc:        Bernd Walter <ticso@mail.cicely.de>, freebsd-hackers@freebsd.org
Subject:   Re: VINUM PANIC ON -STABLE
Message-ID:  <20010916020828.I17960@cicely20.cicely.de>
In-Reply-To: <33343543.1000596877983.JavaMail.imail@doby.excite.com>; from john_wilson100@excite.com on Sat, Sep 15, 2001 at 04:34:32PM -0700
References:  <33343543.1000596877983.JavaMail.imail@doby.excite.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Sat, Sep 15, 2001 at 04:34:32PM -0700, john_wilson100@excite.com wrote:
> 
> On Sat, 15 Sep 2001 21:52:06 +0200, Bernd Walter wrote:
> 
> >  On Sat, Sep 15, 2001 at 12:26:10PM -0700, john_wilson100@excite.com
> wrote:
> >  > I posted a message to -hackers several days ago, complete with a kernel
> >  > backtrace.    The panic is 100% reproducible on my machine running the
> >  > latest -stable.
> >  
> >  I saw two unsualy points in your config.
> >  1. why use more than one vinum partition per physical drive?
> >     vinum itself handles it very well - it's a volume manager.
> 
> If one gets corrupted (for whatever reason), the other one will keep going
> (assuming that one volume will not touch another's partitions)

If you don't trust in vinum not to mix subdisk range...
But I don't think vinum has problems here.
The difference is that you have 2 instances of the vinum configuration
per disk.

> >  2. I don't think that a stripe size not matching n * page size is a
> >     good choice.
> >  Both points shouldn't crash your box but it's at least not very common.
> 
> A stripe size that is a power of 2 will put all superblocks on the same
> subdisk.

The Superblocks are all in the same offset in a cylindergroup.
That means:
1. It's sufficient to enshure that the cylindergroups start at with
   alternating disks.  Well with 279 you don't have to think if it
   meets the requirement.
2. If you loose one disk with 279k stripe size you are not interested
   in reading this superblocks anymore.
   This is a difference with concatenating where you have a good chance
   you recover some of the data and there you don't have to wory about
   this anyway.

> >  > Does anyone care?  My message to Greg Lehey was rejected by his mail
> server,
> >  > but I am pretty sure he reads both -hackers and -stable.
> >  
> >  There must be a reson to reject.
> 
> Let's hear it.

I know that he had an URL explaining what he rejects.
But I'm almost shure that it's described in the error message you
got back.

-- 
B.Walter              COSMO-Project         http://www.cosmo-project.de
ticso@cicely.de         Usergroup           info@cosmo-project.de


To Unsubscribe: send mail to majordomo@FreeBSD.org
with "unsubscribe freebsd-hackers" in the body of the message




Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20010916020828.I17960>