Loopback detected, port went error-disabled suddenly

One working switch port on my Cisco Catalyst 2950 suddenly went down by itself!

Of course, my phone rang when I was having dinner, and then I had to tell my colleagues how to do a port reset step by step. Luckily, after the port reset, the port went back on.

I did not pay much attention to this event, until it happened to the same port again! This time, I drilled in to check the log, and found something suspicious:

.Jan 30 10:12:25: %ETHCNTR-3-HALF_DUX_COLLISION_EXCEED_THRESHOLD:
Half-duplex loopback detected, collision threshold exceeded on FastEthernet0/24.

.Jan 30 10:12:25: %PM-4-ERR_DISABLE:
loopback error detected on Fa0/24, putting Fa0/24 in err-disable state

.Jan 30 10:12:26: %LINEPROTO-5-UPDOWN:
Line protocol on Interface FastEthernet0/24, changed state to down

.Jan 30 10:12:27: %LINK-3-UPDOWN:
Interface FastEthernet0/24, changed state to down

After my survey on Cisco.com, I found a workaround to this: just disable keepalive on that interface:

Switch # config t
Switch (config)# int f0/24
Switch (config-if)# no keepalive

I believe that it is a software caveat. The phrase mentioned on the web page is:

..... The suggested workaround is to disable keepalives and upgrade to Cisco IOS Software Release 12.2SE or later. .....


References:
Errdisable Port State Recovery on the Cisco IOS Platforms