I have recently upgraded a LAN from 2950s to 3750s. Sitting off the 3750s are several 3-com hubs as well as user PCs directly attached. Previously the hubs were connected solely to the 2950s.
The 3750s are reporting the following error frequently:
*Mar 8 11:17:37: %ETHCNTR-3-HALF_DUX_COLLISION_EXCEED_THRESHOLD: Collisions at
FastEthernet1/0/10 exceed threshold. Considered as loop-back.
The 3750 switchports that connect the 3-com hubs are reporting excessive collisions and shutting the port down (err-disable). There is a workaround, but this takes at least 30secs to kick in and reset the port - impacting users:
errdisable recovery cause loopback
errdisable recovery interval 30
The 3750 ports are configured as 10m half duplex for all 3-com port connections, yet still report excessive collisions. This error was not previously seen with the older 2950 switches. Is this a bug or known issue? Is there a better workaround than waiting 30secs each time for the automatic port reset?
switch model: WS-C3750-24TS
switch IOS: c3750-ipbase-mz.122-25.SEE1