OpenVMS Cluster, Fibrechannel and the WWID

[Andrew Rycroft1]

Hi,


I have a cluster of an rx7620 and rx4640 with dual port Qlogic FC cards installed in each. These are connected to a SAN.

If I do a SHOW DEV FG, it shows me the FC port name and FC node name for each card :-

%SYSMAN-I-OUTPUT, command execution on node ITP002
Device FGA0:, device type QLogic ISP23xx FibreChannel, is online, shareable,
error logging is enabled.
Error count 0 Operations completed 95
Owner process "" Owner UIC [SYSTEM]
Owner process ID 00000000 Dev Prot S:RWPL,O:RWPL,G,W
Reference count 0 Default buffer size 0
Current preferred CPU Id 1 Fastpath 1
Current Interrupt CPU Id 1
Document Name: untitled

FC Port Name 5006-0B00-0038-A3A4 FC Node Name 5006-0B00-0038-A3A5
Device FGB0:, device type QLogic ISP23xx FibreChannel, is online, shareable,
error logging is enabled.
Error count 0 Operations completed 95
Owner process "" Owner UIC [SYSTEM]
Owner process ID 00000000 Dev Prot S:RWPL,O:RWPL,G,W
Reference count 0 Default buffer size 0
Current preferred CPU Id 0 Fastpath 1
Current Interrupt CPU Id 0
FC Port Name 5006-0B00-0038-A3A6 FC Node Name 5006-0B00-0038-A3A7
%SYSMAN-I-OUTPUT, command execution on node ITP001
Device FGA0:, device type QLogic ISP23xx FibreChannel, is online, shareable,
error logging is enabled.
Error count 0 Operations completed 95
Owner process "" Owner UIC [SYSTEM]
Owner process ID 00000000 Dev Prot S:RWPL,O:RWPL,G,W
Reference count 0 Default buffer size 0
Current preferred CPU Id 3 Fastpath 1
Current Interrupt CPU Id 3
FC Port Name 5006-0B00-0032-9914 FC Node Name 5006-0B00-0032-9915
Device FGB0:, device type QLogic ISP23xx FibreChannel, is online, shareable,
error logging is enabled.
Error count 0 Operations completed 95
Owner process "" Owner UIC [SYSTEM]
Owner process ID 00000000 Dev Prot S:RWPL,O:RWPL,G,W
Reference count 0 Default buffer size 0
Current preferred CPU Id 2 Fastpath 1
Current Interrupt CPU Id 2
FC Port Name 5006-0B00-0032-9916 FC Node Name 5006-0B00-0032-9917
SYSMAN>


If I do a show device from either systems is displays the fibrechannel disks, and the path to the disks>:-

ITP002>>sh dev d/multi
Device Device Error Current
Name Status Count Paths path
$1$DGA2: (ITP002) Online 0 3/ 3 FGB0.5006-0484-49AE-31F2
$1$DGA131: (ITP002) Online 0 3/ 3 FGB0.5006-0484-49AE-31F2
$1$DGA143: (ITP002) Online 0 3/ 3 FGB0.5006-0484-49AE-31F2
$1$DGA155: (ITP002) Online 0 3/ 3 FGB0.5006-0484-49AE-31F2
$1$DGA167: (ITP002) Online 0 3/ 3 FGB0.5006-0484-49AE-31F2
$1$DGA1371: (ITP002) ShadowSetMember 0 3/ 3 FGB0.5006-0484-49AE-31F2
$1$DGA1389: (ITP002) ShadowSetMember 0 3/ 3 FGA0.5006-0484-49AE-31DD
$1$DGA1407: (ITP002) ShadowSetMember 0 3/ 3 FGA0.5006-0484-49AE-31DD
$1$DGA1425: (ITP002) ShadowSetMember 0 3/ 3 FGA0.5006-0484-49AE-31DD
$1$DGA1443: (ITP002) ShadowSetMember 0 3/ 3 FGA0.5006-0484-49AE-31DD
$1$DGA1449: (ITP002) Mounted 0 3/ 3 FGA0.5006-0484-49AE-31DD
$1$DGA1455: (ITP002) ShadowSetMember 0 3/ 3 FGA0.5006-0484-49AE-31DD
$1$DGA1461: (ITP002) ShadowSetMember 0 3/ 3 FGA0.5006-0484-49AE-31DD

My questions are:-

The WWID of the current paths on both nodes of the cluster are the same, but I can set different paths to access different devices on a per node basis.

Does this mean all the FC traffic for a particular path is going across 1 HBA for the entire cluster?

How come the WWID of the current path in the SHOW device display does not match either the port name or node name of the SHOW DEVICE/FULL FG command?

[Uwe Zessin]

> Does this mean all the FC traffic for a particular path is going across 1 HBA for the entire cluster?

No. Each node will route its own traffic to a single virtual disk via one path. The traffic for different devices can go though different paths.

> does not match either the port name or node name

'SHOW DEVICE/MULTIPATH' displays the WWPNs of the storage controller ports (the targets), while 'SHOW DEVICE/FULL FG' displays the WWPNs and WWNNs of the server's Fibre Channel adapter ports (the initiators).

[John H. Reinhardt]

-- Does this mean all the FC traffic for a particular path is going across 1 HBA for the entire cluster?

The HBA listed in the path is for that node and that device. All the traffic for that path on that node will go through that HBA for that device. It is not a cluster-wide setting. AFAIK you have to manually decide which path you want for each device if you don't want the default path when the system starts up. So for traffic balancing purposes you would need a bit of knowledge about the device usage. Of course, VMS will automatically switch the paths if something should go wrong on any given path and the device become unreachable.

-- How come the WWID of the current path in the SHOW device display does not match either the port name or node name of the SHOW DEVICE/FULL FG command?

The WWID of the current path in the SHOW DEVICE for the disk is that of the storage element not the HBA on each node. In my case (I have a pair of HSG80's) my SHOW DEVICE for the disk shows the WWID for the port on the HSG that the lun is available on - not the WWID of the HBA.

[Robert Atkinson]

Andrew, the attached document may help you to understand how this works a little better.

Rob.

[Wim Van den Wyngaert]

We distribute the paths evenly over the controllers at boot time. So, the 1st and 5th disk gets path 1, the 2nd and 6th path 2, etc.
This because our load changes all the time due to the nature of the applications.

Wim