MSA1000 vs. Linux - Wrong LUN Mapping

[natale titotto]

Hello

We have an MSA1000 SAN Storage with redundant controller connected through an external 2/8 switch with two DL380 G3 servers with one FCA2214 HBA each. Everything has been working fine since a couple of years with Red Hat Linux RHES v.3.
Some months ago we bought a HP DL380 G4 server and equipped it with another identical FCA2214, and I've been trying to use with RH same version, then with RHES v.4 and even with Debian Sid w/ kernel 2.6.15 (not certified, but just to try) and I always have the same problem, no matter which distro or version or kernel I use: the LUN I've created with ACU on the MSA1000 for use with the new server is mapped twice on it, once for each controller:

# cat /proc/scsi/scsi
Attached devices:
Host: scsi0 Channel: 00 Id: 00 Lun: 00
Vendor: COMPAQ Model: MSA1000 Rev: 4.32
Type: RAID ANSI SCSI revision: 04
Host: scsi0 Channel: 00 Id: 00 Lun: 02
Vendor: COMPAQ Model: MSA1000 VOLUME Rev: 4.32
Type: Direct-Access ANSI SCSI revision: 04
Host: scsi0 Channel: 00 Id: 01 Lun: 00
Vendor: COMPAQ Model: MSA1000 Rev: 4.32
Type: RAID ANSI SCSI revision: 04
Host: scsi0 Channel: 00 Id: 01 Lun: 02
Vendor: COMPAQ Model: MSA1000 VOLUME Rev: 4.32
Type: Direct-Access ANSI SCSI revision: 04

I get errors of this kind at boot time:

Device sdb not ready.
end_request: I/O error, dev sdb, sector 0
Buffer I/O error on device sdb, logical block 0

and similar errors trying to initialize the volume (luckily, I guess ;)

The correct situation in the working servers is the following (both controllers are recognized but only the active one maps the volume):

# cat /proc/scsi/scsi
Attached devices:
Host: scsi0 Channel: 00 Id: 00 Lun: 00
Vendor: COMPAQ Model: MSA1000 Rev: 4.32
Type: Unknown ANSI SCSI revision: 04
Host: scsi0 Channel: 00 Id: 00 Lun: 01
Vendor: COMPAQ Model: MSA1000 VOLUME Rev: 4.32
Type: Direct-Access ANSI SCSI revision: 04
Host: scsi0 Channel: 00 Id: 01 Lun: 00
Vendor: COMPAQ Model: MSA1000 Rev: 4.32
Type: Unknown ANSI SCSI revision: 04

Any idea? Where's my fault?
Thank you all
--
natale

[Uwe Zessin]

Both controllers always present a SCSI LUN for each logical disk:

- the active controller so that you can process I/O
- the passive controller so that the multipath software knows that there is a failover path

> Device sdb not ready.

That is a sign that the server has hit the standby paths. That is OK.

--

You have a configuration error:
> scsi0 Channel: 00 Id: **00 Lun: 02
> scsi0 Channel: 00 Id: **01 Lun: 02

a single Fibre Channel adapter (scsi0) must not have access to both MSA controllers. (Id: 00 + 01).

--

If you want redundancy, you need:

- two controllers

- two FC switches or at least one switch with proper zoning, but then this single switch is a SPOF.

- a server with two Fibre Channel adapter ports (either two single-port adapters or one dual-port adapter)

- a piece of multipath software for your operating system
-- it could be Secure Path or
-- the Qlogic adapter driver peroperly configured