You are viewing a plain text version of this content. The canonical link for it is here.
Posted to modperl@perl.apache.org by Perrin Harkins <pe...@elem.com> on 2006/11/15 16:55:00 UTC
[OT] noatime
Most of the apache tuning guides, as well as the MySQL docs, recommend
turning off atime on your filesystem (setting noatime). Are other
people doing this? Do you know if it has helped with performance? Have
you found any problems with it?
- Perrin
Re: [OT] noatime
Posted by Frank Wiles <fr...@wiles.org>.
On Wed, 15 Nov 2006 12:35:38 -0800
"Philip M. Gollucci" <pg...@p6m7g8.com> wrote:
> Perrin Harkins wrote:
> > Most of the apache tuning guides, as well as the MySQL docs,
> > recommend turning off atime on your filesystem (setting noatime).
> > Are other people doing this? Do you know if it has helped with
> > performance? Have you found any problems with it?
> I don't do it with /www file systems, but when I do FBSD build worlds
>
> mount -u -o noatime /usr/src
> mount -u -o async /usr/obj
>
> Thats incredibly stable.
>
> It helps quite substantially with performance.
Another thing I thought I'd mention, you don't have to do it for
an entire filesystem. You can use chattr +A /path/to/file and it
will just not record atime for that file ( or files ).
Nice if you want to play around and see what type of impact it
will have on your system without having to remount your filesystem.
---------------------------------
Frank Wiles <fr...@wiles.org>
http://www.wiles.org
---------------------------------
Re: [OT] noatime
Posted by "Philip M. Gollucci" <pg...@p6m7g8.com>.
Perrin Harkins wrote:
> Most of the apache tuning guides, as well as the MySQL docs, recommend
> turning off atime on your filesystem (setting noatime). Are other
> people doing this? Do you know if it has helped with performance? Have
> you found any problems with it?
I don't do it with /www file systems, but when I do FBSD build worlds
mount -u -o noatime /usr/src
mount -u -o async /usr/obj
Thats incredibly stable.
It helps quite substantially with performance.
--
------------------------------------------------------------------------
Philip M. Gollucci (pgollucci@p6m7g8.com) 323.219.4708
Consultant / http://p6m7g8.net/Resume/resume.shtml
Senior Software Engineer - TicketMaster - http://ticketmaster.com
1024D/A79997FA F357 0FDD 2301 6296 690F 6A47 D55A 7172 A799 97F
When I call your name, Girl, it starts to flame
Burning in my heart, Tearing it all apart..
No matter how I try My love I cannot hide....
Re: [OT] noatime
Posted by Carl Johnstone <mo...@fadetoblack.me.uk>.
> recommend turning off atime on your filesystem (setting noatime).
Done it on both Solaris and Linux boxes, had no problems with either. Didn't
benchmark it, but the servers were really struggling IO-wise at the time and
the change had a noticeable improvement.
Like anything it depends where your bottlenecks are - if you've got 15K RPM
SCSI disks connected to decent SCSI RAID cards slotted onto a 64bit PCI
bus - and your setup only accesses 3 different files, you won't see any/much
of an improvement. If you've got a software RAID on some old IDE disks and a
directory tree of a couple of million files then you'll wonder how you
managed without.
Our current production servers don't have it switched on - we've not noticed
any particular issue IO-wise. Might try it and see if it makes a difference.
Carl
Re: [OT] noatime
Posted by Frank Wiles <fr...@wiles.org>.
On Wed, 15 Nov 2006 10:55:00 -0500
Perrin Harkins <pe...@elem.com> wrote:
> Most of the apache tuning guides, as well as the MySQL docs,
> recommend turning off atime on your filesystem (setting noatime).
> Are other people doing this? Do you know if it has helped with
> performance? Have you found any problems with it?
I do this all of the time, it's not a huge speed gain, but it does
remove the overhead of a filesystem write for every fopen().
I've not run into any problems with this and have used it with
various software ( PostgreSQL, MySQL, Sendmail, mod_perl, vanilla
Apache, etc, etc. ) on both ext2 and ext3.
I doubt it will cause you a problem. Think about this way, when was
the last time "last accessed time" was useful to software you were
writing? :) It's almost always last modified that you need/want.
---------------------------------
Frank Wiles <fr...@wiles.org>
http://www.wiles.org
---------------------------------
Re: [OT] noatime
Posted by "Kevin A. McGrail" <km...@pccc.com>.
It absolutely helps performance. Atime is access time. Otherwise basically
everytime a file is accessed, the atime is updated. This causes a lot of
little writes to the hard drive constantly.
I have recently begun turning it off on our servers in the past 2 years
because in 10 years, I've never looked at the record for any purpose
whatsoever. I'm sure if you need to know the last time the file was
accessed, this would be bad to turn off though. However, for all practical
purposes, the mtime (modified time) is what people want.
regards,
KAM
> Most of the apache tuning guides, as well as the MySQL docs, recommend
> turning off atime on your filesystem (setting noatime). Are other people
> doing this? Do you know if it has helped with performance? Have you
> found any problems with it?