From mboxrd@z Thu Jan 1 00:00:00 1970 From: Tore Anderson Subject: Re: dm-rdac not working? Date: Wed, 29 Aug 2007 09:10:03 +0200 Message-ID: <46D51BCB.6070006@linpro.no> References: <46D2E298.1020702@linpro.no> <1188236026.9504.10.camel@linuxchandra> <46D3188A.3090405@linpro.no> <1188245328.9504.41.camel@linuxchandra> <46D33BAA.8090807@linpro.no> <1188249965.9504.45.camel@linuxchandra> <46D3C099.7080504@linpro.no> <1188326877.12737.1.camel@linuxchandra> Reply-To: device-mapper development Mime-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 7bit Return-path: In-Reply-To: <1188326877.12737.1.camel@linuxchandra> List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , Sender: dm-devel-bounces@redhat.com Errors-To: dm-devel-bounces@redhat.com To: sekharan@us.ibm.com Cc: device-mapper development List-Id: dm-devel.ids * Chandra Seetharaman > Do you see multiple failures like shown below during the device probe > time ? Yes, on all paths to the passive controller I get numerous I/O errors, which happens when the kernel attempts to read in the partition table, when LVM wants to look for PV signatures, and so on. This is the behaviour I expect when using RDAC mode. > Hmm. In my storage it is enabled. That is why I do not see that issue. > May be multipath tools should read and ignore the stderr. If AVT is enabled in your storage, you shouldn't need the RDAC hardware handler. When dm-multipath switches pg and starts sending I/O to the passive paths they should go live automatically, at least they do so for me... Bad thing about AVT is of course that when node X in a cluster boots, the partition table scanning will transfer the volume, making the active controller that e.g. nodes Y and Z was using heavily go passive, disrupting I/O and making everything wobbly until node X has finished its bootup procedure. Regards -- Tore Anderson