From mboxrd@z Thu Jan 1 00:00:00 1970 From: btrfs-devel@arbitraryconstant.com Subject: Re: Btrfs v0.16 released Date: Thu, 7 Aug 2008 13:33:47 -0600 (MDT) Message-ID: <61917.68.145.124.201.1218137627.squirrel@avalon.arbitraryconstant.com> Mime-Version: 1.0 Content-Type: text/plain; charset=US-ASCII To: "linux-btrfs" Return-path: List-ID: > If there is no > alternate mirror, the caller gets EIO and in the case of a failed csum, > the page is zero filled (actually filled with ones so I can find bogus > pages in an oops). You mention there will be a utility to scrub the disks to repair stuff like this, but does it make sense to retry the read in event the it came back from the wrong sector or a bit got flipped somewhere or some other transient error?