From mboxrd@z Thu Jan  1 00:00:00 1970
From: NeilBrown <neilb@suse.de>
Subject: Re: Recovery Optimization?
Date: Sat, 19 Feb 2011 15:54:50 +1100
Message-ID: <20110219155450.4c940485@notabene.brown>
References: <ijn1hs$8nl$1@dough.gmane.org>
Mime-Version: 1.0
Content-Type: text/plain; charset=US-ASCII
Content-Transfer-Encoding: 7bit
Return-path: <linux-raid-owner@vger.kernel.org>
In-Reply-To: <ijn1hs$8nl$1@dough.gmane.org>
Sender: linux-raid-owner@vger.kernel.org
To: Jon Forrest <jlforrest@berkeley.edu>
Cc: linux-raid@vger.kernel.org
List-Id: linux-raid.ids

On Fri, 18 Feb 2011 16:06:43 -0800 Jon Forrest <jlforrest@berkeley.edu> wrote:

> I'm just learning how the md system works so what
> I'm going to say might not be sensible.
> 
> I read that if a disk in a RAID5 set goes bad, and the
> disk is replaced by a new one, that the recovery operation
> takes place blindly. By this I mean that all the stripes
> will be read so that new parity blocks can be written.
> But, there might be stripes that contain only blocks
> that aren't used by the filesystem. Wouldn't it be
> good when doing recovery if some kind of allocation
> map were created so that unused stripes wouldn't be
> restored. I would think that depending on how full
> the disk is that this could save time.
> 
> Is this reasonable?
> 
> Cordially,


http://neil.brown.name/blog/20110216044002#5

NeilBrown