fcmsutil query

[Tim Hempstead]

We are currently having intermittent issues accessing some of our SAN tape decks from one of the vPars on our Superdome.

This manifests itself with both ioscan (-fn) and Netbackup processes hanging (to the extent that they cannot be killed off).

Running "fcmsutil /dev/fcd1 nsstat", where fcd1 is the fibre port which is zoned to see the tape devices, there seems to be a high number of "Rejected rff_id", (in the order of 179000 and this is in the same order of magnitude as the login/logout/request stats).

I have no idea what this rff_id is and even whether it could be contributing to the problem we are seeing but I would like to know if anyone on the forum does know what this is, (and know if its a problem which we should be investigating).

Thanks

Tim

[Florian Heigl (new acc)]

I don't know what the meaning behind the counter is, but You seem to have found Your issue, the value is 0 on all hosts I checked here.

Florian

[Florian Heigl (new acc) · Accepted solution]

Quoting:
"The RFF_ID Name Server request shall be used to record the FCâ 4 Features for a specified TYPE that
are supported by a given Port Identifier.
If the FC-4 TYPE corresponding to the TYPE code specified in the RFF_ID Request CT_IU is not registered,
the Name Server shall reject the registration request and shall set the reason code to â Unable to perform
command requestâ and the reason code explanation to â FC-4 TYPEs not registeredâ ."


Does that help You in any way?

[Tim Hempstead]

Florian,

To be honest it doesn't really help that much as I do not understand the things it is talking about. I am not sure whether this indicates a problem with the server configuration or hardware or a configuration/hardware issue elsewhere on the SAN.

Thanks

Tim

[Florian Heigl (new acc)]

I didn't understand too much of it either, I think it can be translated as 'Patchlevel or SAN Configuration issue'.

You should inquire with (first) HP and the SAN Vendor. Get a specific explanation about the counters meaning, and then ask the SAN Vendor about known issues concerning FC (Port) Nameserver and HP-UX hosts.

Have output of /usr/contrib/bin/show_patches and swlist ready for them, and be prepared for some debugging and fingerpointing time at probably both ends. Always forward both information to both sides, then the root issue should be found, it might well be a typo in a host portname assignment :)