Cisco Support Community
cancel
Showing results for 
Search instead for 
Did you mean: 
cancel
Community Member

FCS Errors Catalyst 2950 Fixed if UTP Cable Pulled/Inserted

Operating 100Mbps Full on both ends. Restart of a controller using National DP83816 Ethernet Chip results in significant FCS and Alignment errors that continue until a change is made. Duplex reported OK at both ends. Pulling UTP cable for less than 1 second fixes the problem. Zero errors in an 8 hour period. The problem also goes away if power cycle or restart the Catalyst 2950. Next restart of the DP83816 controller starts the problem all over again until the cable is pulled and re-inserted.

Problem occurs on all four 2950 switches I tried but does not occur on two other brands of switches I had available. Even upgraded the 2950 to latest firmware. It fails on at least 20 of the DP83816 controllers and various types and lengths of cables.

10 Mbps and auto negotiate work fine.

2 REPLIES
VIP Purple

Re: FCS Errors Catalyst 2950 Fixed if UTP Cable Pulled/Inserted

Hello,

the general idea is that FCS and Alignment errors are caused by cables that either exceed the maximum length or are otherwise faulty; since you are experiencing these problems with different cables, I guess that can be excluded as a possible reason. Can you hook up a sniffer to the switch and monitor one of the ´faulty´ ports ?

There is a Cisco document about Cisco Switch and NIC interoperability, but the DP83816 is not mentioned, you might want to check it anyway:

Troubleshooting Cisco Catalyst Switches to NIC Compatibility Issues

http://www.cisco.com/warp/public/473/46.html#nic_comp

What they usually recommend is to install the lastest driver for the NIC, well I guess that´s obvious...

Regards,

Georg

Community Member

Re: FCS Errors Catalyst 2950 Fixed if UTP Cable Pulled/Inserted

Georg,

Thanks for your input. To me the key element of my problem is that the communication works perfectly if I power up the Controller first but fails only if I restart the controller after the Catalyst 2950 is running. When I am in the failing mode, pulling the Ethernet cable and reinserting it fixes the problem every time. It does not fail again until I reset the controller, even during our temperature testing from 0-60degC. Also, auto negotiate works in all cases and negotiates to 100Mbps Full.

I have the feeling it is something more esoteric, such as adaptive equalization. Either the controller or the 2950 are stuck in some mode at restart and I have been unable to determine which devices is changing when I pull/insert the cable. I am leaning toward the switch being the problem since it gets receive errors. (FCS and Alignment) Looking at the documentation for Ethernet chips, there is very little that changes for transmit. It is either 10 or 100, half or full. However, on the receive side there is a lot going on with active filters attempting to recover data over a wide range of amplitudes and frequency compromised signals. The adaptive filters are used to help extract the data. The Cisco documentation does not talk about this at all. If I knew what the switch did on detection of a new link, it would be helpful.

240
Views
0
Helpful
2
Replies
CreatePlease to create content