MPxIO

[chris.eb...@googlemail.com]

Hi All,

We just bought a new Sun (V445) with a dual ported FC array
(StorageTek 2540). I thought that with the we could load balance
across the controllers, however it is coming up as below with one part
active and one port in standby mode. Is that right? I thought that
this was an active/active device...

Spent all day trying to figure this one out and have got nowhere.

sam:/:# mpathadm show lu /dev/rdsk/
c8t600A0B800038B21100000C204781ADF4d0s2
Logical Unit: /dev/rdsk/c8t600A0B800038B21100000C204781ADF4d0s2
mpath-support: libmpscsi_vhci.so
Vendor: SUN
Product: LCSM100_F
Revision: 0617
Name Type: unknown type
Name: 600a0b800038b21100000c204781adf4
Asymmetric: yes
Current Load Balance: round-robin
Logical Unit Group ID: NA
Auto Failback: on
Auto Probing: NA

Paths:
Initiator Port Name: 2100001b320806d4
Target Port Name: 200200a0b838b1f3
Override Path: NA
Path State: OK
Disabled: no

Initiator Port Name: 2100001b32096230
Target Port Name: 200300a0b838b1f3
Override Path: NA
Path State: OK
Disabled: no

Target Port Groups:
ID: 1
Explicit Failover: yes
Access State: active
Target Ports:
Name: 200200a0b838b1f3
Relative ID: 0

ID: 3
Explicit Failover: yes
Access State: standby
Target Ports:
Name: 200300a0b838b1f3
Relative ID: 0

[George Cebulka]

Hi Chris,
MPXIO require a bit of configuration and a reboot.
First,
in the /kernel/drv/scsi_vhci.conf
You need to enable mpxio. Find the following line:
# mpxio global enable/disable switch: setting mpxio-disable="no" will
activate
# I/O multipathing; setting mpxio-disable="yes" disables this feature (do
# not remove this property).
#
mpxio-disable="yes";
and change mpxio-disable to
mpxio-disable="no";

Set the load balancing to "round-robin"

YOu may also have to change device definitions at the end of the file.

I have included my complete scsi_vhci.conf here. This works on a Hitachi
9990 and V445 with qlogic cards and gives balanced IO across both adapters.

cat /kernel/drv/scsi_vhci.conf
#
# Copyright (c) 2001 by Sun Microsystems, Inc.
# All rights reserved.
#
#pragma ident "@(#)scsi_vhci.conf 1.3 01/08/22 SMI"
#
name="scsi_vhci" class="root";
#
# mpxio global enable/disable switch: setting mpxio-disable="no" will
activate
# I/O multipathing; setting mpxio-disable="yes" disables this feature (do
# not remove this property).
#
mpxio-disable="no";
#
# Load balancing global configuration: setting load-balance="none" will
cause
# all I/O to a given device (which supports multipath I/O) to occur via one
# path. Setting load-balance="round-robin" will cause each path to the
device
# to be used in turn.
#
load-balance="round-robin";
#
# Force load driver to support hotplug activity (do not remove this
property).
#
ddi-forceattach=1;
# Automatic failback configuration
# possible values are auto-failback="enable" or auto-failback="disable"
#
# This functionality requires patch 113039-02 (or higher).
#
auto-failback="enable";
#
# For enabling MPxIO support for 3rd party symmetric device need an
# entry similar to following in this file. Just replace the "SUN SENA"
# part with the Vendor ID/Product ID for the device, exactly as reported by
# Inquiry cmd.
#
# This functionality requires patch 113039-02 (or higher).
#
# device-type-scsi-options-list =
# "SUN SENA", "symmetric-option";
#
# symmetric-option = 0x1000000;
#

Good Luck,
George

[chris.eb...@googlemail.com]

Hi George
thanks for your response. I have checked my /kernel/drv/scsi_vhci.conf
and think that's it is is correct. I enabled with stmsboot and
rebooted. The problem seems to be that it thinks my raid is
assymetric.

chris

[John S.]

On Jan 7, 11:02 pm, "chris.eborn.d...@googlemail.com"

The 2540 is not a true active/active like you are thinking... (there's
a word for it but i can't remember). Basically each LUN is "owned" by
a controller. So LUN1 is assigned to Controller A (with Controller B
as the standby, LUN2 is assigned to Controller B (with Controller A as
standby). So in fact BOTH controllers are in an active state... so
what you are doing is load balancing multiple LUNS across
controllers... not a single LUN across both controllers...

hth,
john

[chris.eb...@googlemail.com]

Hi John,

That helped a lot. I undid the MPxIO stuff and had a poke at it using
good old fashioned /dev/dsk/cxtxdx type paths and realised
what you can and can't do. Will rebuild my filesystems with this in
place. We are stuggling to get the performance of the array up though
-
currently writing at around 100MB/s onto 11 15K SAS drives which we
don't think is really right - but that's another story.
Thanks for your input here - it certainly helped clear up some of our
confusion.

Chris

[mr.bm...@gmail.com]

> currently writing at around 100MB/s onto 11 15K SAS drives which we
> don't think is really right - but that's another story.
> Thanks for your input here - it certainly helped clear up some of our
> confusion.

Chris,

Are you using 4Gb FC Cards? If you aren't then 100MB/s is about
right. I've seen ~150MB/s (with higher spikes) with the 4Gb FC cards
(minimally tuned). If you want faster, you might need to make two+
luns, assign then to opposite controllers and then stripe them with a
volume manager.

--Brett Monroe