Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 3 Mar 2004 13:14:30 +0100
From:      Pawel Malachowski <pawmal-posting@freebsd.lublin.pl>
To:        James Read <james@physicalsegment.com>
Cc:        stable@freebsd.org
Subject:   Re: Using read-only NULLFS leads to panic. gdb output included, easy toreproduce.
Message-ID:  <20040303121430.GB35449@shellma.zin.lublin.pl>
In-Reply-To: <00a501c40171$9e0f6dc0$c000000a@jd2400>
References:  <20040302213936.216CB5F103@shellma.zin.lublin.pl> <00a501c40171$9e0f6dc0$c000000a@jd2400>

next in thread | previous in thread | raw e-mail | index | archive | help
On Wed, Mar 03, 2004 at 10:48:16PM -0000, James Read wrote:

> All I can say is that I've had this happen to me before. 'Me too'
> 
> The ports were mounted in exact the same way, but with rw instead of ro.
> 
> Also a few jails were running at the time, infact 3 were. All had /usr/ports
> mount_null'ed inside there jail.

OT, but it is rather bad idea to have rw mounted /usr/ports shared
beetween hosts/jails. Ports tree can be trojaned from one jail and
everything can be compromised this way.

> After running find from the usual periodic scripts it brought the machine
> down every time it ran that script.

True, my setup was runnig successfully for months. The only change I made
few days ago was to start cron inside jails. :)

> If there is a better way / another way to 'mount_null' /usr/ports (or any
> other mount point for that matter) to other places in/on the filesystem, by
> using NFS or other such things, then speak up! I don't like getting panics
> more then anyone else does ;>

I hope this can be fixed inside nullfs. Tjr@ asked about reproducing this
on 5.x -- it was not possible to panic, so a fix may be already in HEAD.
However, second problem (every process stalls after trying do play with
filesystem) was exposed, see other posts in this thread.


-- 
Paweł Małachowski



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20040303121430.GB35449>