Thanks to all who replied; Michael Wheelock, and especially to Alan Davis
from Compaq.
After talking with Dale Miller from Compaq Storage Support, we figured out
that we could not add the fiber drive with the configuration the we had on
the HSG80. V4.0g will not take a drive past LUN 7. So I upgraded to V5.1 of
the O/S and restored everything on the fiber drive to a local drive. We will
look at reconfiguring the HSG80 configuration to work later. Attached at the
replies that I received and below that is my original question.
==========
Either the SCSI or rz manual pages will explain the V4
device naming. For pure parallel SCSI it is pretty
easy:
rz[b-h]#[b-h]
# - A unit number found from the bus and target numbers
of the device; (bus * 8) + target-id.
SCSI-2 supports each target having some number of logical
units. The first [b-h] of the name, encodes the LUN:
b = LUN 1
c = LUN 2
and so on
Since the base device of "rz#" is for the LUN 0 devices,
you don't need to use 'a' for LUN 0, though either works.
The last [a-h] is the partition number which all disks
support.
This is more complicated with Fibre Channel devices because
they don't use the same addressing scheme as parallel SCSI.
Since V4 only support parallel SCSI naming, a data file of
the emx driver translates unit and WWN to a parallel SCSI
address. This may be described in one of the emx related
manual pages.
================
Mike,
If the kernel has already been recompiled to include the emx
device driver for the KGPSA the rest would be :
The hsg must be in transparent failover and SCSI-2 mode.
Make sure your switch zoning is correct so that your system can
see the hsg80.
On the hsg80 assign the unit number so that it will be visible
to your system. If your system is connected to port 1 of the
hsg80 the unit id must be 1-99, port two connectivity is 100-199.
Set the access path offset on the hsg80 so that the unit id is
no more than path offset + 7. This puts the LUN identifier in the
0-7 range that v4.0x can access. Enable the access path on the unit.
Set the unit id so that unix uses that instead of the long WWN identifier
for the disk.
Reboot the unix system or use scsimgr -scan to make the unit visible
to the system. Each hsg becomes a target on the kgpsa's bus, each unit
on the hsg is a LUN. This means that the disk will show up as
kgpsa scsi bus # == bus
hsg unit # - access path offset = lun (must be in range 0 - 7)
target = depends on discovery order on the SAN
lun = numeric mapped to alpha (0 == a, 1 == b, etc)
device name = rz[lun][( bus * 8 ) + target]c
The manuals recommend assigning unit numbers in groups starting
at boundaries divisible by 10(10,20,30,etc) and to set the
access path offset at the 10 boundaries. An example :
units D30, D31
access patch offset 30
unit ids 30, 31
kgpsa is scsi2 (found in boot messages)
target is 0 (first hsg80 found on the SAN)
D30 =
rz[ 30 - 30 ][ 2 * 8 + 0 ]c
rza16c
D31 =
rz[31 - 30][ 2 * 8 + 0 ]c
rzb16c
Clear as mud?
There are whitepapers on the compaq website that document all
of the configuration issues :
ftp://ftp.compaq.com/pub/products/storageworks/techdoc/raidstorage/EK-SMA30-
AN-B01.PDF
Alan Davis
=================
Hi,
I have been testing 4.0F with some HSG80 disk. These two OS's don't support
multi-path, so you will see several drives added when you add the lun. If
you already see the emulex devices during boot-up (ie. emx0: emx1:) then the
HBA's are live. You can add the lun and do a scsimgr -scan_all (or
optionally scan the bus you need to). This will add the drive devices. You
can then use them as a normal disk. If you do not see the emx devices, you
will need to boot the generic kernel. I am not sure, but I believe the rzh
thing (it was rzb for me) is to do with the fact that the HSG exports the
drives as a single target with multiple luns. Not sure on that, but it is
the same with an hsz80.
Michael Wheelock
============
Hi everyone,
I have a DS10 V4.0g that I am adding a fiber hard drive attach to a HSG80.
The hard drive is on LUN 37, and is a 18gig 15K drive.
Can anyone provide me with some detailed instructions on how to do this?
Will tru64 automatically identify the new hard drive when I reboot with the
generic kernel or is there a better way, so that I dont have to take the
server down?
I noticed on another server that the fiber drives are named as rzh## versus
rz## for the internal drives, where is a link or documentation that defines
all the different devices and there reference codes ie) rz or rzh etc?
Thanks,
Mike Kirkland
Unix System Administrator
National Data Corporation
Phone: (918) 481-2817
Fax: (918) 481-4275
mike.kirkland_at_ndchealth.com <mailto:mike.kirkland_at_ndchealth.com>
NDC®| HEALTH
6100 South Yale Avenue
Suite 1900
Tulsa, OK 74136
Received on Tue Oct 23 2001 - 19:33:40 NZDT