Runt Errors on 6500 Gigabit link

[cmcdougall]

I started getting heavily impacted traffic on a gigabit link between a 6509 and a 6506(Both running SUP2, 8.4(3)). The only problem I could see was that runt errors were increasing on one side of the link only. I shut it down and traffic switched over to the link to our other backbone switch, with traffic running fine. Link negotiation is defined the same on both links. Is this possibly a hardware error and on which end? Sh tech-support shows no anomalies.

[Richard Burts]

Charles

I wonder if there could be a duplex mismatch. If there is a connection and one side is full duplex and the other side is half duplex, there is a possibility that the half duplex side may begin sending, receive data, detect a collusion, stop transmitting, resulting in a runt on the full duplex side if it is within the first 64 bytes of the frame. Another symptom of duplex mismatch would be late collisions on the half duplex side.

HTH

Rick

[cmcdougall]

Rick,

That was my first thought but the config is the same for the bad link as it is for the good link I am using now. There are only errors on the far end of the bad link.

Though I did just notice that there are 14 FCS-ERR errors on the far end of the bad link that I had not noticed before. I don't know exactly when these occurred.

Bad link:

Near end

set trunk 1/2 desirable negotiate 1-1005,1024-4094

Far end(Where I saw runts):

set trunk 5/9 auto negotiate 1-2,101-103,253,800,805,807,1000-1005,2531

Good Link:

Near end:

set trunk 1/1 desirable negotiate 1-1005,1024-4094

Far end(No errors):

set trunk 5/8 auto negotiate 1-2,101-103,253,800,807,1000-1005,2531

Thanks,

Chuck

[Richard Burts]

Chuck

Could you perhaps post the output of show port for each of the ports involved?

HTH

Rick

[cmcdougall]

Rick,

See attached for sh ports.

Chuck

[bjw · Accepted solution]

I've seen this in a case where 50um MMF was driven with LX/LH GBICS between a 6506 and 6513. In our case it was a case of Differential Mode Delay. Changing the GBICS to SX solved the rare issue.

If you are on SMF, them I would check/reseat/changeout the fiber patch cables.

[estiles · Accepted solution]

I agree. I too have seen this when using different gbics. We had a problem very similar to yours. We changed fiber jumper cables and had the fiber runs checked by the cable vendor to be sure the connectors were not dirty or a problem on one side of the link (rx or tx). All is well now after chagning fiber jumper cable.

[cmcdougall]

Thank you everyone for your input.

I tested the fiber path between the switches and found a high loss between 2 of the buildings. I used a different fiber pair. When the link came back up, there were no more errors.

Chuck

[Richard Burts]

Chuck

Thanks for posting back to the forum indicating that the problem was solved (and for the ratings). It makes the forum more useful when people can read about a problem and know that they will also see what the solution to the problem was. I encourage you to continue your participation in the forum.

HTH

Rick