Skip site navigation (1)Skip section navigation (2)
Date:      Fri, 6 Feb 2004 18:22:44 +1100
From:      Peter Jeremy <PeterJeremy@optushome.com.au>
To:        Harald Schmalzbauer <h@schmalzbauer.de>
Cc:        freebsd-current@freebsd.org
Subject:   Re: Fwd: Re: How to calculate bsdlabel size
Message-ID:  <20040206072244.GT908@cirb503493.alcatel.com.au>
In-Reply-To: <200402051908.09836@harrymail>
References:  <200402051908.09836@harrymail>

next in thread | previous in thread | raw e-mail | index | archive | help
On Wednesday 04 February 2004 21:31, Harald Schmalzbauer wrote:
> I manually created disklabels with a size number of 10485760.
> I wanted to have 5 GigaByte big labels, so 5*1024*1024*1024/512=10485760.
> Now after doing a newfs and mounting the new label, df -h reports a size
> of 4.8GB.
> Can someone please enlighten me?

Which df column?  If you're concerned about "missing" blocks, what does
"df -b" or "df -k" report?  What parameters did you use for newfs?

You can expect to lose 3-5% of disk space for metadata overheads.
Each inode taken 256 bytes and each cylinder group takes several FS
blocks.  If you want to fit 5GB of data onto a filesystem, you're
going to need to allocate something like 5.2GB of disk space.

>But it's not my problem. What you mean results in limited available space,
> but doesn't have any influence on Size summary. Btw I did a newfs with -m 0
> so it can't be the reaseon.

This is strongly non-recommended.  The UFS algorithms are designed on
the assumption that there are always free blocks.  When you get below
5-10% free space, the performance will degrade significantly and you
will start getting file fragmentation.

Peter



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?20040206072244.GT908>