Skip site navigation (1)Skip section navigation (2)
Date:      Wed, 10 Jan 2007 12:57:02 +0100 (CET)
From:      Oliver Fromme <olli@lurza.secnetix.de>
To:        freebsd-stable@FreeBSD.ORG, ghelmer@palisadesys.com
Subject:   Re: 6.x loosing record of free space after filesystem fills?
Message-ID:  <200701101157.l0ABv2qq089648@lurza.secnetix.de>
In-Reply-To: <45A3E3B9.4030205@palisadesys.com>

next in thread | previous in thread | raw e-mail | index | archive | help
Guy Helmer wrote:
 > Oliver Fromme wrote:
 > > Why are you using those blocksize and fragsize settings?
 > > (If you store large files, then you should at least also
 > > decrease the inode density, using the -i option.)
 > >   
 > These settings were chosen to optimize I/O throughput for Postgresql on 
 > the theory that a 64KB block size would maximize disk throughput in the 
 > general case (especially for a RAID 10 system) and an 8K frag size would 
 > match Postgresql's page size.

I don't think that that theorie holds true in reality.  Did
you perform any benchmarks to verify it?  In fact, I would
expect the performance to be better when using a block size
of just 8KB and a frag size of 1 KB.

By the way, this is an excerpt from the tuning(7) manpage:

 | FreeBSD performs best when using 8K or 16K file system block
 | sizes.  The default file system block size is 16K, which provides
 | best performance for most applications, with the exception of
 | those that perform random access on large files (such as database
 | server software).  Such applications tend to perform better with
 | a smaller block size, although modern disk characteristics are
 | such that the performance gain from using a smaller block size
 | may not be worth consideration.  Using a block size larger than
 | 16K can cause fragmentation of the buffer cache and lead to
 | lower performance.

Guy Helmer wrote:
 > I wasn't aware of any known regressions in 6.x regarding large 
 > filesystem block sizes...

I'm not aware of any regressions either.  64 KB bsize and
8 KB fsize didn't work reliable in 4.x, and the situation
doesn't seem to have gotten worse (maybe it has gotten
better with UFS2, but I didn't perform extensive tests
with it because the non-standard bsize/fsize pessimize
performance anyway).

Best regards
   Oliver

-- 
Oliver Fromme,  secnetix GmbH & Co. KG, Marktplatz 29, 85567 Grafing
Dienstleistungen mit Schwerpunkt FreeBSD: http://www.secnetix.de/bsd
Any opinions expressed in this message may be personal to the author
and may not necessarily reflect the opinions of secnetix in any way.

"That's what I love about GUIs: They make simple tasks easier,
and complex tasks impossible."
        -- John William Chambless



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?200701101157.l0ABv2qq089648>