From mboxrd@z Thu Jan 1 00:00:00 1970 From: Keith Busch Subject: Re: NVMeoF multi-path setup Date: Thu, 30 Jun 2016 19:14:55 -0400 Message-ID: <20160630231454.GC18890@localhost.localdomain> References: <1467323858.15863.3.camel@ssi> <20160630225207.GB22293@redhat.com> Mime-Version: 1.0 Content-Type: text/plain; charset="us-ascii" Content-Transfer-Encoding: 7bit Return-path: Content-Disposition: inline In-Reply-To: <20160630225207.GB22293@redhat.com> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: dm-devel-bounces@redhat.com Errors-To: dm-devel-bounces@redhat.com To: Mike Snitzer Cc: Ming Lin , device-mapper development , linux-nvme@lists.infradead.org List-Id: dm-devel.ids On Thu, Jun 30, 2016 at 06:52:07PM -0400, Mike Snitzer wrote: > AFAIK, hch had Intel disable that by default in the hopes of avoiding > people having dm-multipath "just work" with NVMeoF. (Makes me wonder > what other unpleasant unilateral decisions were made because some > non-existant NVMe specific multipath capabilities would be forthcoming > but I digress). For the record, Intel was okay with making SCSI a separate config option, but I was pretty clear about our wish to let it default to 'Y', which didn't happen. :) To be fair, NVMe's SCSI translation is a bit of a kludge, and we have better ways to get device identification now. Specifically, the block device provides 'ATTR{wwid}' available to all NVMe namespaces in existing kernel releases. From mboxrd@z Thu Jan 1 00:00:00 1970 From: keith.busch@intel.com (Keith Busch) Date: Thu, 30 Jun 2016 19:14:55 -0400 Subject: NVMeoF multi-path setup In-Reply-To: <20160630225207.GB22293@redhat.com> References: <1467323858.15863.3.camel@ssi> <20160630225207.GB22293@redhat.com> Message-ID: <20160630231454.GC18890@localhost.localdomain> On Thu, Jun 30, 2016@06:52:07PM -0400, Mike Snitzer wrote: > AFAIK, hch had Intel disable that by default in the hopes of avoiding > people having dm-multipath "just work" with NVMeoF. (Makes me wonder > what other unpleasant unilateral decisions were made because some > non-existant NVMe specific multipath capabilities would be forthcoming > but I digress). For the record, Intel was okay with making SCSI a separate config option, but I was pretty clear about our wish to let it default to 'Y', which didn't happen. :) To be fair, NVMe's SCSI translation is a bit of a kludge, and we have better ways to get device identification now. Specifically, the block device provides 'ATTR{wwid}' available to all NVMe namespaces in existing kernel releases.