[Top] [All Lists]

Re: unexpected high fragmentation, any ideas?

To: Marc Lehmann <schmorp@xxxxxxxxxx>
Subject: Re: unexpected high fragmentation, any ideas?
From: Nathan Scott <nathans@xxxxxxx>
Date: Mon, 4 Apr 2005 15:04:13 +1000
Cc: Steve Lord <lord@xxxxxxx>, Russell Cattelan <cattelan@xxxxxxx>, Chris Wedgwood <cw@xxxxxxxx>, linux-xfs@xxxxxxxxxxx
In-reply-to: <20050403215230.GA919@xxxxxxxxxx>
References: <20050403173940.GA12673@xxxxxxxxxxxxxxxxxxxxx> <20050403004653.GA981@xxxxxxxxxx> <20050403050542.GB5727@xxxxxxxxxxxxxxxxxxxxx> <20050403135805.GC24559@xxxxxxxxxx> <42500F94.1070603@xxxxxxxxxxx> <20050403004653.GA981@xxxxxxxxxx> <20050403050542.GB5727@xxxxxxxxxxxxxxxxxxxxx> <20050403135805.GC24559@xxxxxxxxxx> <4250000C.5070709@xxxxxxx> <20050403215230.GA919@xxxxxxxxxx>
Sender: linux-xfs-bounce@xxxxxxxxxxx
User-agent: Mutt/1.5.3i
On Sun, Apr 03, 2005 at 11:52:30PM +0200, Marc Lehmann wrote:
> > The reverse ordering you notice is a little odd, it is definitely
> > supposed to prefer going the other way.

This could be the effect of being called from a memory reclaim
path rather than the regular pdflush path.

> > > ext3 looks much better as it (seemingly) tries to allocate the files
> > > in different block groups when multiple files are being written.
> > 
> > we could put a sysctl or something mount option --- im not sure on the
> > whole it's useful

Certainly sounds useful, at the very least as an option, if not
by default.  IIRC, Dave Chinner did some experiments awhile back
along these lines, and came up with the same conclusion, but not
sure he's coded up a solution there yet.

> What could be useful would be a mount option to force the block allocator
> to treats files in the same directory like files in different directories.

Or perhaps a per-directory flag/attribute.

> Would probably be easier to hack into mythtv, will think about that

Right.  You can experiment with xfs_io, which knows how to issue
that command (as well as how to do buffered vs. direct, etc, etc).

> > makes a differnce as that behaviour is something that could be
> > tweaked/changed as a mount option maybe
> Would it be easier to just disable it in the kernel? Patching kernel code
> would be certainly easier for me than implementing it inside mythtv. I'd
> look myself, but do you have any pointers at which file to look? That
> would be great, thanks!
> On Sun, Apr 03, 2005 at 12:03:27PM -0700, Chris Wedgwood <cw@xxxxxxxx> wrote:
> > for really slow writes i found a large biosize helped.  i've had this
> Is this somewthing inside xfs or is this just setting the st_blksize stat

This is inside XFS, it sets the amount preallocated beyond EOF for
buffered file writes.  Something along these lines will probably
be merged soon, as we're seeing needs for it on at least two other



<Prev in Thread] Current Thread [Next in Thread>