From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: X-Spam-Checker-Version: SpamAssassin 3.4.0 (2014-02-07) on aws-us-west-2-korg-lkml-1.web.codeaurora.org Received: from vger.kernel.org (vger.kernel.org [23.128.96.18]) by smtp.lore.kernel.org (Postfix) with ESMTP id 42CF1C433EF for ; Sun, 22 May 2022 13:00:37 +0000 (UTC) Received: (majordomo@vger.kernel.org) by vger.kernel.org via listexpand id S239017AbiEVNAg (ORCPT ); Sun, 22 May 2022 09:00:36 -0400 Received: from lindbergh.monkeyblade.net ([23.128.96.19]:54876 "EHLO lindbergh.monkeyblade.net" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S237612AbiEVNAf (ORCPT ); Sun, 22 May 2022 09:00:35 -0400 Received: from verein.lst.de (verein.lst.de [213.95.11.211]) by lindbergh.monkeyblade.net (Postfix) with ESMTPS id B914315FD4; Sun, 22 May 2022 06:00:33 -0700 (PDT) Received: by verein.lst.de (Postfix, from userid 2407) id 51BD568AFE; Sun, 22 May 2022 15:00:30 +0200 (CEST) Date: Sun, 22 May 2022 15:00:29 +0200 From: Christoph Hellwig To: Anand Jain Cc: Qu Wenruo , Christoph Hellwig , fstests@vger.kernel.org, linux-btrfs@vger.kernel.org Subject: Re: [PATCH 2/2] btrfs: test repair with corrupted sectors interleaved over multiple mirrors Message-ID: <20220522130029.GA25364@lst.de> References: <20220520164743.4023665-1-hch@lst.de> <20220520164743.4023665-3-hch@lst.de> <791e365d-eb41-9073-80b7-40ee9a42d659@oracle.com> <1fc36e3f-8fa9-2bfb-ede1-d4f852bcb8cf@oracle.com> MIME-Version: 1.0 Content-Type: text/plain; charset=us-ascii Content-Disposition: inline In-Reply-To: <1fc36e3f-8fa9-2bfb-ede1-d4f852bcb8cf@oracle.com> User-Agent: Mutt/1.5.17 (2007-11-01) Precedence: bulk List-ID: X-Mailing-List: fstests@vger.kernel.org On Sun, May 22, 2022 at 06:26:26PM +0530, Anand Jain wrote: > Agreed this method is unreliable. But there is no other choice. > Unless we integrated [1] patch in the ML. > > [1] btrfs: introduce new read_policy device > > [1] is more reliable. You can set which mirrored device to read. I actually thought about something like this as the reliable way to hit a mirror. Besides testing I could also thing of some other use cases for it where you want to avoid a mirror by default for some reason (e.g. because it is network attached store while the other one is local).