I have a WS-C3750G-12S-S switch which is used as a backup for critical switches that have multiple fibre links. In the event the primary fibre link to the 6500 fails, the data will pass through this 3750 to the 6500 at our 2nd site until the link is restored. So for 99.9% of the time little or no traffic will cross these links.
The fibre is primarily multi-mode with 1 single-mode link to our second site.
I was having a look at the logs for this switch (attached) and noticed that for a few months we have been getting Line Protcol UPDOWN errors on ports 1,2,5 and 7.
These ports don't have anything in common, different switches, different fibre etc. There is also no commonality between the switches that are unaffected.
Despite these errors occurring, we had a recent planned outage of our primary fibre link between our two sites which caused all inter-site traffic to cross the single-mode fibre link for 3 hours without incident.
To fix the problem I tried upgrading to the latest IOS image and power-cycling the switch, but the errors returned within a couple of minutes of it coming back online.
I tried swapping over some of the fibre patch leads and GBICS but this also did not fix the problem and in a couple of cases the link failed to come online at all, even when I swapped them back.
At this stage the only common thread in this problem was the switch, so I replaced it with a spare 3750 we had using the same GBICS and leads, immediately all of the links were restored and the problem disappeared. I brought the old switch back to my office, erased the config and connected 2 spare switches to it using spare fibre patch leads and GBICs, I used multi-mode on port 1 and single-mode on port 2, the links were trunked and I left them running for a few hours. When I returned I saw the same error messages appear in the logs which pretty much confirms a fault of some kind with the switch.
There were no POST errors.
So my questions are, can I stop these errors from occurring? If I can't fix it, can Cisco fix it without it being prohibitively expensive? If it can't be fixed should I just ignore the errors? During our fibre outage the switch seemed to function OK, so I could keep this as a spare in case another 3750 failed completely.
Any advice appreciated.