fcpdev vs. fcparray

[Peter Heinemann]

We're running 6 rp8400s (12 hosts) connected to an EMC symm. OS is hpux 11-11.

On one system (only) I can see only 8 emc disks. ioscan shows one diffence, in the driver for the ext_bus.
On the ok systems:

fc 0 0/0/8/0/0 td CLAIMED INTERFACE HP Tachyon TL/TS Fibre Channel
fcp 0 0/0/8/0/0.1 fcp CLAIMED INTERFACE FCP Domain
ext_bus 5 0/0/8/0/0.1.31.0.0 fcparray CLAIMED INTERFACE FCP Array Interface
target 8 0/0/8/0/0.1.31.0.0.0 tgt CLAIMED DEVICE
disk 4 0/0/8/0/0.1.31.0.0.0.0 sdisk CLAIMED DEVICE EMC SYMMETRIX

on the problem system:

fc 1 1/0/10/0/0 td CLAIMED INTERFACE HP Tachyon TL/TS Fibre Channel
fcp 1 1/0/10/0/0.97 fcp CLAIMED INTERFACE FCP Domain
ext_bus 5 1/0/10/0/0.97.26.255.1 fcpdev CLAIMED INTERFACE FCP Device Interface
target 9 1/0/10/0/0.97.26.255.1.3 tgt CLAIMED DEVICE
disk 11 1/0/10/0/0.97.26.255.1.3.0 sdisk CLAIMED DEVICE EMC SYMMETRIX
disk 12 1/0/10/0/0.97.26.255.1.3.1 sdisk CLAIMED DEVICE EMC SYMMETRIX

I've done insf -e and ioscan, rebooted, rmsf for the problem disks and insf -e and ioscan again; no change.

The difference in effect of the fcpdev vs. fcparray drivers seems evident; but how did the fcpdev driver get established, and how do I switch it?

Is this EMC turf or HP-UX turf?

Thanks.

[Eugeny Brychkov]

Peter, in first example device seen in Volume Set Addressing mode (fabric), but in the second in Pripheral Addressing mode (public loop).
I guess if you have switches then some devices are working in NL (public loop, second example) mode? I feel that these modes should be tuned at the EMC side, not at HPUX's
Eugeny

[Gerrit Beyken]

Hi Peter,

Eugeny is right.
You can check the topology of your adapter by doing "fcmsutil /dev/tdx" where Dev/tdx stands for the devicefile of your tachyon-adapter.
Most probably you are attached over a switch to your emc. you can also see the topology when you telnet your switch and do a switchshow (for brocades i think this is the command).
Maybe this port is set to loop.
You can modify this but keep in mind that diskdevices would change on that HW-path.

Best regards

Gerrit

[Peter Heinemann]

Thanks for the replies. Although not exactly the problem, they did help me convince the San administrator to look further at the configuration. There were various incorrect settings.

[Peter Hurwood]

I know this was a while ago but you said that: "Although not exactly the problem, they did help me convince the San administrator to look further at the configuration. There were various incorrect settings."

I have a similar problem now deploying a CLARiiON array in a SAN with HP-UX 11.11 boxes. Do you remember what switch tweaks were required to make this work?

thanks

Peter

[RAJESH GANGADHARAN]

Hi,

Any body got an answer for the first question,
Peter, it will be greatful if you can share the resolution for your probelm, as I am facing a similar problem.

[Uwe Zessin]

I am not an HP-UX expert (I can barely spell it ;-) , but:
VSA vs. PDLA - isn't that defined as the 'host mode' or 'operating system selection' for each fibre channel adapter on the storage array? I think this is about how to address LUNs on a target, not whether fabric/loop, no?