xfs
[Top] [All Lists]

Re: howto preallocate to minimize fragmentation

To: Stewart Smith <stewart@xxxxxxxxxxxxxxxx>
Subject: Re: howto preallocate to minimize fragmentation
From: Nathan Scott <nathans@xxxxxxx>
Date: Thu, 13 Apr 2006 15:23:29 +1000
Cc: Eric Sandeen <sandeen@xxxxxxx>, Ying-Hung Chen <ying@xxxxxxxxxxxxxx>, Andrew Ho <andrewho@xxxxxxxxxxxxx>, linux-xfs@xxxxxxxxxxx
In-reply-to: <1144905000.9181.32.camel@localhost.localdomain>; from stewart@flamingspork.com on Thu, Apr 13, 2006 at 03:10:00PM +1000
References: <43329839.2070005@yingternet.com> <4332A22B.6070708@sgi.com> <4332BFCC.8050803@yingternet.com> <4332C248.70503@sgi.com> <4332C636.9070509@yingternet.com> <4332CE65.2000500@animezone.org> <4332CF04.2060604@yingternet.com> <4332D17E.6060608@sgi.com> <1144905000.9181.32.camel@localhost.localdomain>
Sender: linux-xfs-bounce@xxxxxxxxxxx
User-agent: Mutt/1.2.5i
On Thu, Apr 13, 2006 at 03:10:00PM +1000, Stewart Smith wrote:
> On Thu, 2005-09-22 at 10:45 -0500, Eric Sandeen wrote:
> > something like this:
> >          err = xfsctl(argv[1], fd, XFS_IOC_RESVSP64, &fl);
> 
> Are we going to see this be part of posix_fallocate in glibc any time
> soon?

I'd be dumbstruck if so.  It really needs a proper fs-independent
system call.  This keeps coming up, and an interface has even been
discussed on LKML in the past - just needs some guy to go do it.
Some guy called "Stewart" maybe. ;)

> The patch would be fairly trivial and then let application
> programmers use a standard interface that has a nice fallback in case of
> non-XFS file systems.

The patch to MySQL would be just as trivial, surely. ;)

cheers.

-- 
Nathan


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