From owner-freebsd-scsi@FreeBSD.ORG Mon Dec 21 11:20:55 2009 Return-Path: Delivered-To: freebsd-scsi@freebsd.org Received: from mx1.freebsd.org (mx1.freebsd.org [IPv6:2001:4f8:fff6::34]) by hub.freebsd.org (Postfix) with ESMTP id C5C7F1065676 for ; Mon, 21 Dec 2009 11:20:55 +0000 (UTC) (envelope-from danny@cs.huji.ac.il) Received: from kabab.cs.huji.ac.il (kabab.cs.huji.ac.il [132.65.16.84]) by mx1.freebsd.org (Postfix) with ESMTP id 80B358FC1D for ; Mon, 21 Dec 2009 11:20:55 +0000 (UTC) Received: from pampa.cs.huji.ac.il ([132.65.80.32]) by kabab.cs.huji.ac.il with esmtp id 1NMgJp-000587-VZ; Mon, 21 Dec 2009 13:20:53 +0200 X-Mailer: exmh version 2.7.2 01/07/2005 with nmh-1.2 To: Miroslav Lachman <000.fbsd@quip.cz> Mime-Version: 1.0 Content-Type: text/plain; charset=us-ascii Date: Mon, 21 Dec 2009 13:20:53 +0200 From: Daniel Braniss Message-ID: Cc: freebsd-scsi@freebsd.org Subject: Re: iSCSI initiator and Dell PowerVault MD3000i X-BeenThere: freebsd-scsi@freebsd.org X-Mailman-Version: 2.1.5 Precedence: list List-Id: SCSI subsystem List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Mon, 21 Dec 2009 11:20:55 -0000 > Daniel Braniss wrote: > > > >> So instead of iSCSI connected storage with multipath failover, I got > >> singlepath storage causing system freeze in case of network disconnection :( > > > > well, if you unplug the net, or reboot the target, it will recover. > > it's when you down/up the network when problems occur, and I'm > > looking into it. I took a machine, with 2 nics, connected to 2 vlans, A is the default, B is on the same vlan as my target iscsi server. if I unplug or do an 'ifconfig netB down', the iscsi_initiator will recover - after about 1 minute - reconnecting to the target via netA > Just the note to my tests with multipathing. I asked about MD3000i one > friend who is aware of DELL arrays and he told me that it is > 'active/passive' / 'dual active' type of array and Virtual Disks (LUNs) > are always accessible only by one controller. In case of failure of > currently active controller path, the driver must send some command to > the other controller to take ownership of the LUN to this controller. > > I searched the net for some more information and it seems that it is > handled by RDAC on Linux > (http://fxr.watson.org/fxr/source/drivers/md/dm-mpath-rdac.c?v=linux-2.6) but > we don't have it in FreeBSD? > http://lists.freebsd.org/pipermail/freebsd-scsi/2009-December/004045.html > > Do you have some knowledge about this? Is it something that can be done > in iSCSI or is it something for another part / layer in the FreeBSD? > Let me see if I can untangle all these acronyms :-) iSCSI is scsi over ip, it's 'usually' presented some 'scsi device', tape, disk, via some target. It's the target that controls the real hardware, so a hardware error at the raid controller level cannot be handled by the initiator (so no RDAC), specially when the target provides a virtual disk (which is about 100% true for all the targets I have encountered so far). A RAID controller is not a SCSI device, and so it can't be controlled by the iSCSI initiator. Multipath, in iSCSI it means more than one network path to the same target/lun, the current iscsi_initiator does not handle this, but since istg handles this, I'm looking into it. it would be helpful if you can describe what you are trying to accomplish. danny