From mboxrd@z Thu Jan 1 00:00:00 1970 From: Christoph Hellwig Subject: Re: [PATCH 0/5] target/rbd: distributed lun reset support Date: Tue, 5 May 2015 23:48:12 -0700 Message-ID: <20150506064812.GA5181@infradead.org> References: <1430819778-24483-1-git-send-email-mchristi@redhat.com> <20150505174939.GA21297@infradead.org> <55499D46.9040008@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Return-path: Content-Disposition: inline In-Reply-To: <55499D46.9040008@redhat.com> Sender: target-devel-owner@vger.kernel.org To: Mike Christie Cc: Christoph Hellwig , ceph-devel@vger.kernel.org, target-devel@vger.kernel.org List-Id: ceph-devel.vger.kernel.org On Tue, May 05, 2015 at 11:49:10PM -0500, Mike Christie wrote: > On 05/05/2015 12:49 PM, Christoph Hellwig wrote: > > Can you explain why you want to hook into the lun reset code? I know > > the PR code needs to issue implicit task aborts, but I don't > > understand the need to do something with lun resets. > > I thought if you exported a LU through multiple LIO nodes and its ALUA > state was active optimized/non-optimized on those nodes, then you would > have to do similar cleanup across the cluster. For example, don't we > have to fail commands on those other nodes, and then also depending on > the TAS bit value return task aborted statuses? Oh, you're actually implementing active/active exports. Yes, in that case we'll need to distribute LUN resets. > > As for the API: I would much prefer to have things at the block layer > > in some form than adding side band protocol for specific functionality. > > > > Ok. For what you need for your NFS cluster work, will you have a > request_queue or block_device or gendisk? All of them. It's just another consumer of the in-kernel block devices interface.