We have Nexus 5020s running 5.1(3) on our server Access Tier which show excessive Input Discards on several interfaces going up to the Distribution 7000s, but also on several of our interfaces connecting to the servers. This is in the magnitude of 120,000 discards in 80 minutes and 30,000 discards/80 minutes on two uplinks into the Dist Tier, but also 12,000/80 Min, 8,000/80 Min on several ports connecting to servers. I understand that this could be indicative of Head-of-Line-Blocking (switch egress congestion causing backup on the switch ingress), so when attempting to address the large number of input discards on the interfaces facing the Dist Tier, I'm looking for signs of egress congestion going to the servers, IE: high TX Interface Load, High output errors, etc. However, I do not see ANY signs of egress congestion.
The below was a great doc, but I think I need to look elsewhere.
http://www.sysadmintutorials.com/cisco-nexus-5500-input-packet-discard-troubleshooting/
Does anyone have any ideas, thoughts, insights or suggestions on where else to look?
Thanks!
Mike