Cisco Support Community
cancel
Showing results for 
Search instead for 
Did you mean: 
Announcements

Welcome to Cisco Support Community. We would love to have your feedback.

For an introduction to the new site, click here. If you'd prefer to explore, try our test area to get started. And see here for current known issues.

New Member

How to you stop outdiscards from occurring on ethernet ports?

I have a number of switches being monitored that are reports many outdiscards on various ports. What is the easiest way to prevent outdiscards from occurring on ethernet ports? 

Regards, Mark
  • LAN Switching and Routing
5 REPLIES
Hall of Fame Super Gold

Discards on an ethernet port

Discards on an ethernet port normally happens when the downstream device constantly keeps sending a "pause" frame to the switch.  The switch then sends the intended packets into a buffer pool.  When this buffer gets full, the switch will then discard the oldest packet.  This is when discard counter increments.  

 

A few ways to manage discards is QoS and another is check the configuration of the downstream device.  

 

One of the most common causes of discards is when your switch has a very low buffer or memory pool and it couldn't handle micro-burst of data traffic.  This happens when you have a Catalyst switch and you've got high-speed VM as downstream clients.

New Member

Leo, Taking your statement

Leo, 

Taking your statement into account, what is a possible solution I could look at implementing on my side to resolve the outdiscards?

Regards, Mark
Hall of Fame Super Gold

There are several methods to

There are several methods to determine if QoS (or the lack of) is the cause of your discards.  

My favourite is the quick-and-very-dirty method of setting the port to 10/100 Mbps (interface command:  speed auto 10 100) and see if the discards are still incrementing or not.  

New Member

Most of the ports

Most of the ports experiencing the outdiscards are already configured for auto negotiation. 

Regards, Mark
New Member

This maybe a negotiation

This maybe a negotiation problem with your specific hosts...... Currently I have an issue with incrementing discards on a bunch of switch ports running 1gb auto, I have a Gigamon aggregator switch tapping these problem switch ports sending each hosts data to a traffic analyzer tool. I can see over 1s period the max bandwidth is 600mb for most hosts (all 1gb links) but if I dig deeper over 1ms and 10ms microbursts I can see the link almost constantly hitting 1GB, ie.- hitting the max. QOS is really not going to make much difference here with my type of traffic, the solution is to go to a 10Gb solution. Not an easy solution, this costs money! How many packets per/sec per 5 min input and output are you seeing? Mine are approx 10,000-15,000 on average on a WS-X6748-GE-TX card. sh int gi1/10

643
Views
0
Helpful
5
Replies
This widget could not be displayed.