Hello,
try and use the exec command:
clear tcp tcb
There is a bug, but it (in theory) applies only to the 12.2S train.
I found the additional explanation below:
These thrashing checks are intended to determine if a process is, for some
reason, not doing its job. The thrashing check on watched queues (which is
the troublesome message which is signaling) checks the number of elements
on the queue. If this number remains the same for a given number of
schedulings, the message is printed.
Unfortunately, some queues are length-limited, meaning that if the router
gets very busy, the queues always stay at the maximum. As a result, the
thrashing code in the scheduler gets confused and thinks that these queues
have not been handled. The thrashing code has determined that the process
which was supposed to handle the queue was not doing its job and prints
the thrashing message.
The scheduler has been changed in later Cisco IOS software code. To keep
track of whether the queues have been changed (so it can better determine
whether or not the process is thrashing), the scheduler now notes whenever
an item is removed from the queue, and only prints the thrashing message
if nothing gets removed for a while.
Most of the time, the queue thrashing message is simply cosmetic.
These messages are not always caused by a software bug. They may be issued
in response to either instantaneous or sustained demand on the router.
Increased or persistent messages may indicate that the traffic load needs
to be reviewed.
