[c-nsp] Catalyst 2970 and third-party SFPs

[Adam Rothschild]

Hello,

Does anybody have experience using third-party SFPs with the Cisco
Catalyst 2970 series of switches?

I've tried connecting an assortment of Finisar85 0nm (SX) SFPs
(identical to the ones Cisco _sells_, possible exception being the
serial numbers :-) to a brand new WS-C2970G-24TS-E.

As would be expected, the ports came up in errdisable state, and the
following error messages were dumped to syslog:

  00:01:21: %GBIC_SECURITY_CRYPT-4-VN_DATA_CRC_ERROR: GBIC in port 65561 has bad crc
  00:01:21: %GBIC_SECURITY_CRYPT-4-VN_DATA_CRC_ERROR: GBIC in port 65562 has bad crc
  00:01:21: %GBIC_SECURITY_CRYPT-4-VN_DATA_CRC_ERROR: GBIC in port 65563 has bad crc
  00:01:21: %GBIC_SECURITY_CRYPT-4-VN_DATA_CRC_ERROR: GBIC in port 65564 has bad crc

So, I added "no errdisable detect cause gbic-invalid" to my
configuration, and bounced the ports.  Even tried bouncing the ports;
they continued to report a down/down state:

  Switch#sh int g0/25
  GigabitEthernet0/25 is down, line protocol is down (notconnect)

I'd suspect a vendor lock mechanism, but have had luck on with
Catalyst 6500 and 3750 platforms with the above methodology.
Additionally, Cisco states on their web site[1]:

  "Cisco's industry-standard SFP is a hot-swappable input/output device
  that plugs into a Gigabit Ethernet port/slot, linking the port with
  the fiber-optic network. SFPs can be used and interchanged on a wide
  variety of Cisco products and can be intermixed in combinations of
  IEEE 802.3z- compliant 1000BaseSX, 1000BaseLX/LH, or 1000BaseZX
  interfaces on a port-by-port basis."

...which would seem inconsistent with such a lock at a minimum, and
possibly bordering on false advertising, if this is indeed the case.
So, I'm hoping that's not it.

Anybody out there who's run into this issue before, or is otherwise in
a position to shed some light on the situation?

Thanks,
-a

[1] <http://www.cisco.com/en/US/products/hw/modules/ps5000/ps5248/index.html>