Troubleshooting Performance Forced vs Autonegotiate Switched Port

[jwmock]

Users sitting behind a 16 port 10/100 Netgear switch report slow response or the inability to save documents to a file server attached to a 4006. The Netgear switch is attached to a port that was set forced 100 full. Pings timed out as the packet size increased and the switch port showed FCS, runts, etc. errors. We changed the switch port to autonegotiate and it reported back as a-100 and a-full. All the errors on the port went away and the performance improved greatly. What is the difference between forcing 100/full and having the switch report back that it autonegotiated 100/full? I thought the recommendation was to force if you knew both ends would support it.

[skarundi]

The description you posted matches the classic symptoms of a duplex mismatch.

Carefully review the document on cisco.com concerning this issue.

Title:

Troubleshooting Cisco Catalyst Switches to NIC Compatibility Issues

Doc ID: 17053

URL: http://www.cisco.com/en/US/products/hw/switches/ps700/products_tech_note09186a00800a7af0.shtml#auto_neg_valid

[glen.grant]

In order for auto to work it must be set to auto on both the Netgear switch and the 4006 . Seeing that you had hardcoded the 4006 and the the netgear box was trying to auto and could not , the default for the netgear box will then fall back to 100/half thus you have a duplex mismatch and you will get all kinds of errors and slow performance . If you had hardcoded the 4006 to 100/half it would have worked ok though not as good having it auto at 100/full .