Recently, we have set up Network Performance Monitor and it is showing a high amount of transmit discards on every interface of one of our switches.
AnnexStack is actually two Catalyst 3750 switches - one is a 24 port (WS-C3750-24PS-S) and the other is a 48 port (WS-C3750-48PS-S). They are connected via stackwise cable, and the 24 port is being elected as the stack master. These switches are both running Version 12.2(44)SE5. I have searched and found cases of people having a similar issue where the discards are reported incorrectly. However, the "show platform port-asic stats drop" is showing a large amount of TxQueue drops as well, so I do not believe it is being reported incorrectly.
From what I have read, transmit discards can be a sign of network congestion. It may be important to note that 70 of the 72 ports available are in use, but we have not noticed any disruption of service or problems relating to this issue. All of the traffic is trunked through the GigabitEthernet2/0/1 port back to our core switch via fiber.
We plan to upgrade soon to two 48 port switches stacked in the same configuration, but we are worried that the amount of transmit discards may cause more trouble with 24 additional ports (if the issue is related to network congestion).
Any ideas or insight as to what could be causing this, things to test, etc, would be much appreciated.
Thank you,
Dan