From mboxrd@z Thu Jan 1 00:00:00 1970 From: Josh Durgin Subject: Re: ceph and scsi reservation-like locking Date: Mon, 30 Dec 2013 13:43:14 -0800 Message-ID: <52C1E8F2.6060107@inktank.com> References: <6035A0D088A63A46850C3988ED045A4B68567948@BITCOM1.int.sbss.com.au> Mime-Version: 1.0 Content-Type: text/plain; charset=ISO-8859-1; format=flowed Content-Transfer-Encoding: 7bit Return-path: Received: from mail-gg0-f170.google.com ([209.85.161.170]:56659 "EHLO mail-gg0-f170.google.com" rhost-flags-OK-OK-OK-OK) by vger.kernel.org with ESMTP id S932246Ab3L3VoE (ORCPT ); Mon, 30 Dec 2013 16:44:04 -0500 Received: by mail-gg0-f170.google.com with SMTP id 24so2405333gge.29 for ; Mon, 30 Dec 2013 13:44:03 -0800 (PST) In-Reply-To: <6035A0D088A63A46850C3988ED045A4B68567948@BITCOM1.int.sbss.com.au> Sender: ceph-devel-owner@vger.kernel.org List-ID: To: James Harper , "ceph-devel@vger.kernel.org" On 12/28/2013 06:34 PM, James Harper wrote: > Is the rbd locking feature-compatible with scsi3 persistent reservations? It wasn't designed with scsi-3 exactly in mind, but using exclusive locks and fencing might suffice. To fence a client, you'd get its address from rbd_list_lockers() and blacklist it with the 'ceph osd blacklist add' command, which can be done via rados_mon_command() these days. Blacklisting adds an entry in the osdmap so osds return an error when blacklisted clients try to access them. By default a blacklist entry expires after 1 hour. > To use ceph as a storage backend for a Hyper-V cluster directly (rather than through an iscsi gateway), it looks like I'll need a virtual scsi3 device that supports persistent reservations. I'm not sure exactly what parts of persistent reservations Hyper-V relies on - let us know how it goes! Josh