I am currently deploying a Cisco UCS mini solution with two FI 6324 modules. I have set the first two ports in FC mode and configured the FI modules in FC switch mode. This is a smaller solution with an EMC VNX connected directly to the FI modules. I have followed the guides and I have successfully configured all connectivity.
I have however noticed that if there is a sync or link loss on the FC ports, the ports report an error and the port stays down. This could for example be a firmware upgrade on the storage system which will cause a brief link down status. Another example is to simply replace a cable, causing a short time of link down. I still have redundant paths, meaning I do not get a service disruption as such, but the only solution to bring back full redundancy is to manually disable and then re-enable the FC port in the FI module.
I don't have a lot of experience with Cisco as a whole, but I assume this behaviour could be by design. The FI module detects an excessive number of errors and disables the port. Is this assumption correct? If so, can I configure the port error threshold or the number of attempts to re-establish connection? I would like to automate the operations as much as possible.
Not sure what version of Mini you are running, but there are some known defects with UCS Mini and EMC storage:
Give these a glance and see if any of the conditions match yours. This is probably what you are experiencing.
Let me know what further questions you have.
Thanks for the links, I have reviewed the suggestions.
However, setting the Fill Word (or Fill Pattern) to IDLE I think would indirectly mean I would have to set the FC link to 4 Gbps, which is not at all acceptable.
The ARB(FF) Fill Word is the recommended (or even only official correct way, not sure) according to the T11 committe when running at 8 Gbps.
It is also the default as far as I could tell in the Cisco UCS mini.
We just now upgraded firmware to I think the latest 3.0(2e), and we still have a problem with the ports coming up, although now the ports stays in an "initializing" state, instead of "loss of sync".
I have set the corresponding VNX ports to 8 Gbps instead of auto.
A bounce of the port will bring up the link, so we're not completely down but I of course would like to avoid this step.