can a bad port damage ISL SFPs?

[Turnor Wootton]

We have an odd problem here and I am wondering if anyone has encountered anything similar. We had an SFP on an HP p-class blade switch (brocade)slowly go bad and finally fail completely. It was an ISL to one of our core switches (Brocade 4100). Replacing the SFP did not bring the link up. Replaced the fiber with no success. Replacing the (lights green) SFP in the core switch brought the link up. Swapped back old fiber. Everything ok.

A few months later, the same port in the blade switch exhibits the same symptoms. Slow failure resulting in users continuing to hit it, since it hadn't failed. Failed completely during troubleshooting. Swapped out both SFPs (on the ISL) this time on general principles. Problem went away.


The problem just occured again. Same port on the blade switch. Being able to occaisonally learn from mistakes, I dove in and found the errors and disabled the port which solved the immediate problem. Blade switch was completely replaced. Reusing exisiting SFPs on the assumption that it was a port problem (same port every time). Problem port did not come back on-line, until I replaced the SFPs on both sides of the ISL..


Does anyone know what could be causing this? Is there a known problem with switch ports damaging SFPs on both sides of an ISL link? I really have done fairly good PD here. Swap one out, swap the other out, swap out fiber, send supportsaves to HP (support contract) for both switches, etc.


I'm puzzled. Thanks, Doug

[Amar_Joshi]

Just wondering whether you tried to re-insert the same SFP on blade and core switch or not? even before doing that did you try disable/enable switch and blade ports, would that be making any difference to any of the status?

Also, do you see anything "interesting" in portshow output? It's better if you have 2 outputs (when everything is GOOD and when everything goes BAD) to compare the port statistics and see what's going wrong over the time.

It will be interesting to see what kind (model#) of SPFs you are using at each end. Personally I don't believe in the theory that one SFP can damage other, unless they are using different wavelengths.

I believe, you would have done check on latest firmwares release notes too.

[Turnor Wootton]

As I said, I'm puzzled. I discovered the problem port based on porterrshow output (pre-failure) the errors on the port were on the order of 1000x the other ports. While in testing it went to no-light and failed completely.

While in the error state, disabling and renabling the port did not help (errors continued after re-enabling). After replacing the switch, the SFP (from the old switch) was "still" bad. Replacing only the "Bad" SFP did not help. I had to replace both sides of the ISL. I did not try to reboot the core switch.

I am running current (6.x) firmware with Finisar SFPs.

I did not try and disable/re-enable the port on the core side, just the one on the edge side.

-Doug