EMC Clarion CX3 Storage

[Kevin Raven (UK)]

We have several OpenVMS clusters that boot and use EMC DX series storage.

I have a remote team who are trying to set up a two node OpenVMS Alpha ES45 cluster.

They are using EMC clarion CX3 storage for the boot disk and cluster storage.

So far they can boot both the ES45 from the OpenVMS 8.2 CD and see all the required disks (18 spindles) as $1$DGA999: to $1$DGA1017: .
All disk have two path to disk.
Via wwwidmgr we can see both the HBA adapters in each server and from a show dev dga at the SRM prompt see the disk.
So based on the fact that the system disk will be unit 999 , they are trying to run the required quickset command and it does not return a boot path.

The question is ...does anyone else in VMS land boot from a EMC clarion CX3 ?
sho dev dga/multi
%PMA0 CPU00: 13-OCT-2009 15:52:00 Port is Reinitializing. Check the Error Log.
Device Device Error Current
Name Status Count Paths path
$1$DGA999: () Online 0 2/ 2 PGA0.5006-0162-41E0-93E6
$1$DGA1000: () Online 0 2/ 2 PGA0.5006-0162-41E0-93E6
$1$DGA1001: () Online 0 2/ 2 PGA0.5006-0162-41E0-93E6
$1$DGA1002: () Online 0 2/ 2 PGA0.5006-0162-41E0-93E6
$1$DGA1003: () Online 0 2/ 2 PGA0.5006-0162-41E0-93E6
$1$DGA1004: () Online 0 2/ 2 PGA0.5006-0162-41E0-93E6
$1$DGA1005: () Online 0 2/ 2 PGA0.5006-0162-41E0-93E6
$1$DGA1006: () Online 0 2/ 2 PGA0.5006-0162-41E0-93E6
$1$DGA1007: () Online 0 2/ 2 PGA0.5006-0162-41E0-93E6
$1$DGA1008: () Online 0 2/ 2 PGA0.5006-0162-41E0-93E6
$1$DGA1009: () Online 0 2/ 2 PGA0.5006-0162-41E0-93E6
$1$DGA1010: () Online 0 2/ 2 PGA0.5006-0162-41E0-93E6
$1$DGA1011: () Online 0 2/ 2 PGA0.5006-0162-41E0-93E6
$1$DGA1012: () Online 0 2/ 2 PGA0.5006-0162-41E0-93E6
$1$DGA1013: () Online 0 2/ 2 PGA0.5006-0162-41E0-93E6
$1$DGA1014: () Online 0 2/ 2 PGA0.5006-0162-41E0-93E6
$1$DGA1015: () Online 0 2/ 2 PGA0.5006-0162-41E0-93E6
$1$DGA1016: () Online 0 2/ 2 PGA0.5006-0162-41E0-93E6
$1$DGA1017: () Online 0 2/ 2 PGA0.5006-0162-41E0-93E6
$$$
P00>>init
Initializing...


OpenVMS PALcode V1.98-43, Tru64 UNIX PALcode V1.92-33

*** keyboard not plugged in...
starting console on CPU 0
initialized idle PCB
initializing semaphores
initializing heap
initial heap 240c0
memory low limit = 204000 heap = 240c0, 17fc0
initializing driver structures
initializing idle process PID
initializing file system
initializing timer data structures
lowering IPL
CPU 0 speed is 1250 MHz
create dead_eater
create poll
create timer
create powerup
access NVRAM
2048 MB of System Memory
Testing Memory
...
probe I/O subsystem
Hose 0 - PCI bus running at 33Mhz
entering idle loop
Hose 1 - PCI bus running at 66Mhz
Hose 2 - PCI bus running at 66Mhz
Hose 3 - PCI bus running at 33Mhz
starting drivers
initializing GCT/FRU at 226000
Initializing dqa dqb eia eib eic eid pga pga0.0.0.1.2 - Nvram read failed.
pgb pgb0.0.0.101.2 - Nvram read failed.
open fibre pgb0.0.0.101.2
pgc pgc0.0.0.2.2 - Nvram read failed.
pgd pgd0.0.0.102.2 - Nvram read failed.
open fibre pgd0.0.0.102.2
pka pkb
Memory Testing and Configuration Status
Array Size Base Address Intlv Mode
--------- ---------- ---------------- ----------
0 1024Mb 0000000000000000 1-Way
1 1024Mb 0000000040000000 1-Way

2048 MB of System Memory
Testing the System
pgb0.0.0.101.2 Link is down.
Testing the Disks (read only)
Testing the Network
pgd0.0.0.102.2 Link is down.
AlphaServer ES45 Console V7.3-2, built on Feb 27 2007 at 13:09:18
P00>>wwidmgr -sho adapter
pgb0.0.0.101.2 Link is down.
pgd0.0.0.102.2 Link is down.
item adapter WWN Cur. Topo Next Topo
pga0.0.0.1.2 - Nvram read failed.
[ 0] pga0.0.0.1.2 2000-0000-c972-324c FABRIC UNAVAIL
pgb0.0.0.101.2 - Nvram read failed.
[ 1] pgb0.0.0.101.2 2000-0000-c972-324b FABRIC UNAVAIL
pgc0.0.0.2.2 - Nvram read failed.
[ 2] pgc0.0.0.2.2 2000-0000-c972-ccd0 FABRIC UNAVAIL
pgd0.0.0.102.2 - Nvram read failed.
[ 3] pgd0.0.0.102.2 2000-0000-c972-cccf FABRIC UNAVAIL
[9999] All of the above.
P00>>sho dev dga
dga13353.13.0.1.2 $0$DGA13353 DGC RAID 0 0326
dga14950.13.0.1.2 $0$DGA14950 DGC RAID 0 0326
dga20968.13.0.1.2 $0$DGA20968 DGC RAID 0 0326
dga24206.13.0.1.2 $0$DGA24206 DGC RAID 0 0326
dga25508.13.0.1.2 $0$DGA25508 DGC RAID 0 0326
dga25928.13.0.1.2 $0$DGA25928 DGC RAID 0 0326
dga25963.13.0.1.2 $0$DGA25963 DGC RAID 0 0326
dga28592.13.0.1.2 $0$DGA28592 DGC RAID 0 0326
dga28727.13.0.1.2 $0$DGA28727 DGC RAID 0 0326
dga28917.13.0.1.2 $0$DGA28917 DGC RAID 0 0326
dga29374.13.0.1.2 $0$DGA29374 DGC RAID 0 0326
dga29502.13.0.1.2 $0$DGA29502 DGC RAID 0 0326
dga30273.13.0.1.2 $0$DGA30273 DGC RAID 0 0326
dga30720.13.0.1.2 $0$DGA30720 DGC RAID 0 0326
dga31108.13.0.1.2 $0$DGA31108 DGC RAID 0 0326
dga461.13.0.1.2 $0$DGA461 DGC RAID 0 0326
dga6395.13.0.1.2 $0$DGA6395 DGC RAID 0 0326
dga792.13.0.1.2 $0$DGA792 DGC RAID 0 0326
dga8222.13.0.1.2 $0$DGA8222 DGC RAID 0 0326
P00>>

From above we have 2 * 2 port HBA cards with fibre in one port on each ...thus why pgb and pgd are shown as link down ....

Thanks
Kevin

[Volker Halle · Accepted solution]

Kevin,

the 'Nvram read failed message' indicates, that the following needs to be issued:

>>> wwidmgr -set adapter -item n -topo fabric

Volker.

[EdgarZamora_1]

You need to correct the nvram read failed errors using wwidmgr -set adapter command. set the topology to fabric.

[EdgarZamora_1]

Oops didn't see Volker beat me to it :)

[Ian Miller.]

VMS V8.2 is sooo 2005 and expected to drop out of standard support when V8.4 comes out next year. I don't expect HP will support booting from EMC drives even if it does happen to work.

[Volker Halle]

Kevin,

if you issue a >>> SHOW DEV DG command on the SRM console AFTER running wwidmgr and BEFORE an INIT, you'll always be confused by random device names.

Try >>> wwidmgr -show wwid

after fixing the Nvram read failed errors.

Only after setting the boot devices and another >>> INIT will >>> SHOW DEV DG command show useful information - hopefully.

Volker.

[Peter Zeiszler]

Before running wwidmgr - do a "show dev dg".
Show w*
This should show devices that are defined with wwid numbers.

show n*
this should show the WW name or port of the cards.

wwidmgr -show wwid
wwidmgr -show reach
wwidmgr -show wwid -item (#) -full
(Should see the multiple paths to the device)

On some of my devices I had to set them up:
wwidmgr -set

wwidmgr -set wwid -item (#) -unit (UDID number)
The first 4 I believe are what you want for the system disk or dump_dev devices. This should set them up as devices in the boot variable wwid0 - wwid3. These should also be the devices that show up when you do a "show device dg" prior to running wwidmgr.

We don't boot off our EMC - but we do boot off XP1024. The other thing we had to make sure of on the EMC is that both systems were accessing the disks through the name port path. The paths on the EMC should be A or B. Otherwise it could show up on the EMC as trespassing.

[Carson Hovey]

EMC has a document which might be of interest: "EMC Host Connectivity Guide for OpenVMS".

I've attempted to attach it here; but it is over 1MB so it doesn't work; but customers should have access to it through EMC PowerLink (see the emc website).

There are minimum levels of CLARiiON FLARE code required for base attach as well as booting from the SAN.