BL20p causing port errors on my switches

[Thomas Kemp_1]

Hello -

I am seeing the ports on my 3900s and 12000s switch from No_sync to Port_Flt only from BL20p blades using the Qlogic FCA2214 mezzanine card; all other systems and storage are running with no errors.

The HBAs originally had firmware 1.33, and I upgraded this to 1.34, but the faults did not go away. The BIOS is enabled on the cards.

The switch ports are set to AN. I am seeing the errors on FOS 4.2.0b, and 4.1.2b

Is there something else that has to be done to stop these errors?

Thanks.

[Steven Clementi]

Did you try to hard set the ports to F/2Gb?




Steven

[Thomas Kemp_1]

Steve -

I seem to recall that I did, but it didn't help.

[Steven Clementi]

How do you have your mezz cards zoned? Are they all together in one zone with your storage? or are you zoning by host?


Steven

[Thomas Kemp_1]

Steve -

We use host oriented zoning. We have one port from the HBA going into one fabric, and the port going into a second fabric, should one fabric or switch fail. The zones consist of one hba with its corresponding EVA 5000 ports.

By the way, these errors seem to occur if the blades are booted, or powered up, but not booted.

Thanks,

[xela]

Hello Thomas,

we had the same problems with same Enviroment.
HBA FW 1.33 and Switch Firmware 4.1.0 or 4.1.1 !
Plz try out "Hard Zoning" - SwitchPort Zoning.
First i had WWN Zoning with use of aliases then i switched to switch port zoning without aliases and the problem was fixed.

Let us know if these fixed your problem :-)

[ianvt]

Has anybody found a cure for this problem? I'm also getting switch port state offline and faulty errors when the server with a FCA2214 HBA is rebooted.
Ian

[Danny Loria_1]

I am having a similar problem. Brocade 12000 with Brocade 3900, HP EVA5000, and BL20p. However, when I create single initiator zoning, I am getting even worse results; existing hosts in thier own zones lose LUN's, hosts with secondary storage only loose acces to tape drives, in other words all of my zoning seems to go nuts. I was concentrating on the switches and zoning as the root cause, but thanks to this thread I will turn my attention to the BL20p's.
However, I do not recommend port-based zoning. Although I will be trying this out when I get onsite tommorrow, it is NOT the HP or Brocade best practice. The best practice is to use Aliases with the WWN, and single initiator zones. If this is not working then HP and/or Brocade should be made aware in order to fix the issue or change the best practices, but, I do not want to put myself in a position were we are running in an unsupported configuration

[Thomas Kemp_1]

As for the BL20p errors on a reboot, or for an idle blade:

When you turn off the blade via iLO or reboot, there is still light being emitted from the HBA or blade enclosure. In not so technical terms, when the tx from the blade / enclosure loses the intelligence of the HBA from a power off or reboot, the switch faults. I've spoken ad nauseum to HP about this, and the only fix I've found, especially for idle blades, is to leave them at fast!util HAB configuration until you're ready to use them.