3 Replies Latest reply on Aug 30, 2012 11:12 AM by joopv

    Packet drops / discards on interfaces


      I'm looking for a way to read the packet drops on the interfaces of cisco switches.


      There is an report-error info table job type, and it contains a lot of port variables.  But no packet drops or discards.


      We have a Cisco 6500 chassis with 6148 48-port 10/100/1000  blades, but since these ports are 1:8 overbooked, we really need to monitor the packet drops to see what the impact of the overbooking is.




        • Re: Packet drops / discards on interfaces

          Hello Joop,


          You could use the Rerport.SNMP.System summary Activity and have it request the following OIDs for each interface:


          Interface Incoming discards:

          Interface outgoing discards:


          Note: You need to add the interface ID to the end of the above OIDs, which means you would have to enter these for every interface.


          In reality, you are not using the correct application to satisfy your need as you are using a network management application to perform the job of a network monitoring application.  My suggestion would be to look at Toolset's NPM tool (http://www.solarwinds.com/products/toolsets/network-performance-monitoring.aspx) or our Orion NPM ( Network Performance Monitor | NPM | SolarWinds) product.  These two applications monitor discards and errors out-of-the-box.


          Hopefully this helps.




          Chris Foley • SolarWinds • Technical Support

          Office Hours: Mon-Fri 8AM-5PM EST 866.530.8040


          explore our IT management solutions for:

          networks|applications|storage|virtualization|log & event

            • Re: Packet drops / discards on interfaces

              Thanks for your prompt reply Chris,


              I think that would be a bit impactical, since the OID numbers range from 1 to 450 on this 6513 chassis (checked with Getif) and the snmp job only has place for 5 OID's.  There are 6 48-port blades (300 physical interfaces) plus a multitude of vlan's, port-channels etcetera.  The table output from the report.error.info would be ideal for this purpose.


              A once-a-week table would be good enough to see if there is packet loss.


              Is there any chance to get these variables added to the report-error-info job.  It seems to me it should be there anyway...