Tachyon Fiber card error

[Nyck_1]

Over the weekend one of my RX4640's lost contact with our SAN. I checked the logs and its coming back with the following error:-

The Fibre Channel Driver received an interrupt from the
adapter indicating an Elastic Store Error Storm
Frame Manager Status Register = 0xa30004d0

I have replaced the cable, GBIC and tried a different port on the switch and still no luck.

Anyone got any other suggestions?

I have a nasty feeling that its the fiber card thats failing:-(

[Ganesan R]

Hi,

Check the card status by fcmsutil.

For example,

#fcmsutil /dev/td0

[Nyck_1]

fcmsutil /dev/td0

Vendor ID is = 0x00103c
Device ID is = 0x001029
XL2 Chip Revision No is = 2.3
PCI Sub-system Vendor ID is = 0x00103c
PCI Sub-system ID is = 0x00128c
Previous Topology = UNINITIALIZED
Link Speed = UNINITIALIZED
Local N_Port_id is = 0x000000
Local Loop_id is = 126
N_Port Node World Wide Name = 0x50060b0000bdbe23
N_Port Port World Wide Name = 0x50060b0000bdbe22
Driver state = AWAITING_LINK_UP
Hardware Path is = 0/5/1/0
Number of Assisted IOs = 0
Number of Active Login Sessions = 0
Dino Present on Card = NO
Maximum Frame Size = 2048
Driver Version = @(#) libtd.a HP Fibre Channel Tachyon XL2 Driver B.11.23.0512 $Date: 2005/09/20 12:22:47 $Revision: r11.23/1

[Steven E. Protter]

Shalom,

Your nasty feeling is probably correct.

Check the physical media, lights in the card fcmsutil and prepare for a hardware service call.

SEP

[Nyck_1]

Hello,

Yup no lights on the card or where its connected at the SAN switch and I have just discovered that we do not have hardware support on this server!!!!!!

Not amused what so ever with this situation.

So will be restoring 250GB to a different server until I can get this issues resolved.

[Matti_Kurkela]

Driver state = AWAITING_LINK_UP
usually means the card is not seeing any incoming light. Whether this is because there is no light coming in or because the card has gone blind, remains to be seen.

Can you physically confirm that light is coming on the fibre from the switch and that the lit fibre is plugged into the RX connector? If all this is true, it looks like the card has failed and requires service.

If the card has a pluggable GBIC, try swapping that first. If the card has an integrated GBIC, you'll have to swap the card.

The optical components in some fibre cards and GBICs tend to run fairly hot. That causes them to age and eventually fail.

Always use at least two fibre cards with production servers! Even the alternate path configuration of a basic HP-UX 11.23 and below will protect you from card failures. 11.31 has built-in fibre load balancing, which is even better. If your OS/HW combination supports OLAR/OLRAD, you can even swap the failed card without shutting down the OS.

MK

[Jozef_Novak]

Hello,

from the fcmsutil output it's clear that the card is not logged into the fabric. So if you tried a different port and cable and GBIC and whatever else, then the card itself is likely at fault.

Is this server connected to the SAN only via this single card ? (not recommended for production data). Looks like you have to replace it.

J.

[Nyck_1]

Hello,

Yup I have tried a different fiber cable, different port and different GBIC and still no luck.

I did point out to the powers that be that we need two fiber cards in each server but they did not want to spend the extra money, typical eh!

[Nyck_1]

Out of interest how straight forward would it be to present the LUNS from the server with the failed fiber card onto a different server?

The SAN I have is a Clariion CX300

[Jozef_Novak]

Hello (again),

I think this should be fairly simple. Add masking to those LUNs for the new server (WWNs of the HBA(s)) and then import the affected volume groups. See vgimport(1m).

J.