Single Path vs. QLogic Failover

[ALBERTO ROCHA_1]

I've two BL20p G3 attached to an MSA1000 w/ 12 mirrored disks (6 on channel A + 6 on channel B) w/ redundant controller, running RHEL4-U4 x86_64 w/ PSP 7.60 drivers.

If i select "SinglePath" when configuring HBA it has a strange behavior: on both nodes, 'fdisk' reports /dev/sdg-h-i-j-k-l and not /dev/sda-b-c-d-e-f

If i select "QLogicFailover", 'fdisk' simply doesn't load /dev/sdX, despite 'lssd' command from 'fibreutils' reports on both nodes:
sda 0,0,0,1 COMPAQ MSA1000 VOLUME 4.48
sdb 0,0,0,2 COMPAQ MSA1000 VOLUME 4.48
sdc 0,0,0,3 COMPAQ MSA1000 VOLUME 4.48
sdd 0,0,0,4 COMPAQ MSA1000 VOLUME 4.48
sde 0,0,0,5 COMPAQ MSA1000 VOLUME 4.48
sdf 0,0,0,6 COMPAQ MSA1000 VOLUME 4.48

First beginners question: once i have a redundant controller and last drivers, i should use "QLogicFailover", right?!

Second question: should i see /dev/sda,b,c,d,e,f for the first node end /dev/sdg,h,i,j,k,l for the second node or should they be equal?

Thx

[Ivan Ferreira]

First beginners question: once i have a redundant controller and last drivers, i should use "QLogicFailover", right?!

Yes, you should use qlogic failover for hosts with multiple redundant HBAs.

Second question: should i see /dev/sda,b,c,d,e,f for the first node end /dev/sdg,h,i,j,k,l for the second node or should they be equal?

You cannot warranty that would be equal but could be. It depends of the SCSI devices attached and the discovery order. Normally, if the hardware is very similar (internal disks, and so) the device files can be the same in the servers.

[ALBERTO ROCHA_1]

Some more hits about the configuration:

i) The contents of '/etc/modprobe.conf' that confirms QLogicFailover driver:

options qla2xxx ql2xmaxqdepth=16 qlport_down_retry=30 ql2xloginretrycount=16 ql2xfailover=1 ql2xlbType=1 ql2xautorestore=0x80
remove qla2xxx /sbin/modprobe -r --first-time --ignore-remove qla2xxx && { /sbin/modprobe -r --ignore-remove qla2xxx_conf; }

resulting '/etc/hp_qla2x00.conf':
qdepth = 16
port_down_retry_count = 30
login_retry_count = 16
failover = 1
load_balancing = 1
excludemodel = 0x0
auto_restore = 0x80

ii) The contents of '/var/log/messages' i can allways see entrys like:
kernel: Buffer I/O error on device sdX, logical block 0

The thruth is that 'fdisk' only see partitions at '/dev/cciss/c0d0' and none at '/dev/cciss/c0d1'

Thx