Skip site navigation (1)Skip section navigation (2)
Date:      Thu, 14 Nov 2013 14:49:58 -0600
From:      Mark Felder <feld@FreeBSD.org>
To:        freebsd-fs@freebsd.org
Subject:   Re: Defaults in 10.0 ZFS through bsdinstall
Message-ID:  <1384462198.13183.47596065.6F8E7BCD@webmail.messagingengine.com>
In-Reply-To: <3D3332FA-0ABF-4573-8E65-4E7FBB37100B@fisglobal.com>
References:  <20131114173423.GA21761@blazingdot.com> <59A9B68B-4134-4217-83F3-B99759174EFE@fisglobal.com> <5285148E.6020903@allanjude.com> <3D3332FA-0ABF-4573-8E65-4E7FBB37100B@fisglobal.com>

next in thread | previous in thread | raw e-mail | index | archive | help
On Thu, Nov 14, 2013, at 12:35, Teske, Devin wrote:
> 
> I have never heard a good argument for having atime on. The performance
> penalty on ZFS is quite large, and it also makes your snapshots grow
> constant. If you have a use for it, you can turn it on I guess. This
> would be solved by having the dataset editor we're planning for 10.1
> 

POLA and POSIX, even though it was a bad decision to invent atime :-)
We've never turned atime off before and it would be a huge surprise to
me, so I'd avocate that we let the admins who know what they're doing
turn it off. I know many Linux distros install with noatime and/or
nodiratime, but I'm 99% sure tools don't create filesystems with atime
flagged to be off by default (tune2fs -O noatime). 

We don't even do installs on UFS with atime disabled by default in fstab
so why should we so suddenly change course for ZFS?

-my 2c



Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?1384462198.13183.47596065.6F8E7BCD>