From mboxrd@z Thu Jan 1 00:00:00 1970 From: James Bottomley Subject: Re: [PATCH] scsi: do not print 'reservation conflict' for TEST UNIT READY Date: Tue, 13 Sep 2016 07:04:56 -0700 Message-ID: <1473775496.2214.6.camel@HansenPartnership.com> References: <1473668453-115818-1-git-send-email-hare@suse.de> Mime-Version: 1.0 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: 7bit Return-path: Received: from bedivere.hansenpartnership.com ([66.63.167.143]:59868 "EHLO bedivere.hansenpartnership.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S1751567AbcIMOFA (ORCPT ); Tue, 13 Sep 2016 10:05:00 -0400 In-Reply-To: <1473668453-115818-1-git-send-email-hare@suse.de> Sender: linux-scsi-owner@vger.kernel.org List-Id: linux-scsi@vger.kernel.org To: Hannes Reinecke , "Martin K. Petersen" Cc: Christoph Hellwig , linux-scsi@vger.kernel.org, Hannes Reinecke On Mon, 2016-09-12 at 10:20 +0200, Hannes Reinecke wrote: > SPC-2 and SPC-3 (or later) differ in the handling of reservation > conflict for TEST UNIT READY. SPC-2 will return 'reservation > conflict', whereas SPC-3 will return GOOD status. > On a mixed system with both SPC-2 and SPC-3 targets one will > see lots of 'reservation conflict' messages from the SPC-2 system but > no messages from the SPC-3 system when eg multipath path checkers. > These messages might confuse the unsuspecting user although in fact > they just signal normal operation. I don't agree with this: a SCSI-2 device will not get properly configured if it's reserved by something else, so you get other strange artifacts of this condition. > So we should not be printing out 'reservation conflict' for > TEST UNIT READY responses. This doesn't sound like a good rationale to me. The way I see it, if this message is actually useful, people would like to see it when they get a reservation conflict. That does mean even when SCSI-2 reservations give one where SCSI-3 would not. The other reason is that it tells you why your device didn't get configured properly: both Test Unit Ready and Read Capacity get conflicts with SCSI-2, whereas they do with SPC-3+ devices (trying to forget SPC-2). You could argue that the entire message needs removing, since it's reporting stuff that mostly only shows when systems using reservations correctly are in operation. James