Date: Tue, 28 Oct 2003 18:23:30 -0500 From: Scott W <wegster@mindcore.net> To: Wes Peters <wes@softweyr.com> Cc: wes@freebsd.org Subject: Re: newfs by fstab directory name? Message-ID: <3F9EFA72.20508@mindcore.net> In-Reply-To: <200310281512.07942.wes@softweyr.com> References: <200310271326.23562.wes@softweyr.com> <20031028120520.A20792@hub.freebsd.org> <200310281512.07942.wes@softweyr.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Wes Peters wrote:
>On Tuesday 28 October 2003 12:05, David O'Brien wrote:
>
>
>>On Mon, Oct 27, 2003 at 01:26:23PM -0800, Wes Peters wrote:
>>
>>
>>>At work we do a lot of dynamic filesystem creation, so we added the
>>>ability to specify the 'special file' argument to newfs via the
>>>fstab mount point directory. Please see the attached patch. If
>>>nobody objects, I'll commit this in a couple of days.
>>>
>>>
>>Not objecting, but I don't follow how the change is to be used.
>>Can you post an example?
>>
>>
>
>Sure. Example from /etc/fstab:
>
>/dev/ad0s1d /tmp ufs rw 2 2
>/dev/ad0s1f /usr ufs rw 2 2
>/dev/ad0s1e /var ufs rw 2 2
>/dev/acd0 /cdrom cd9660 ro,noauto 0 0
>/dev/da0s1e /spool ufs rw,noauto 0 0
>
>The disk space on /spool is managed by the "application" and isn't
>guaranteed to be on-line or even existent when the "system" portion
>loads and starts the application. This space is entirely transient
>data that doesn't need to be saved across reboots. When the
>application starts, it checks to see if /spool is clean; if so it
>just mounts it, if not it newfs's it and then mounts it. This
>space isn't necessarily always "da0s1e" but it is always "/spool"
>across different hardware platforms. We prefer to:
>
> newfs /spool
>
>rather than
>
> . {some file full of shell variables describing the hardware}
> newfs $SPOOL_PARTITION
>
>because the former is slightly more concise. We had a local patch to do
>this in our 4.x code base, but it seemed a general enough change that
>others might find it useful as well. I recall ecountering this same
>problem at DoBox so it appears to be a general problem for disk-based
>appliances, at least if you want to support differing hardware.
>
>
>
This would also be useful for anyone doing any sort of benchmarking
using data sets- I did a code port of LADDIS/SFS to Linux ages ago to do
some NFS/SMB fileserver testing, and I can say, quite a LOT of entirely
temporary data is generated...
Scott
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?3F9EFA72.20508>