4500-26 to 5500G-EI Packet Discards

[kjones1]

I have a new building where I have configured several 4500-26 ports to connect to a centralised pair of 5500G-EI’s in an XRN fabric but the 4500-26 currently have only one link back to the 5500’s (no redundancy). I am using the same XFP GBic’s in the 5500 and the 4500 switches and on a trial version of IMC I am seeing a high number of packets discard on interface/x/y/z.



I find this very strange as this is happening on all receive ports of the 5500 and not in the opposite direction. Given that they are using the same GBic’s, then I would not expect that the sensitivity of the GBic would be the issue. All ports on all switches have been removed from the default VLan on all switches. Edge ports are then connected to the 4500 as access ports with the main link tagged down to the 5500 with each having its own VLAN ID. The 5500 is then configured with all these VLan ID’s and acts as a routed connection between these VLan’s and the VLan that also connects back to our HQ.



The users say that the network is great there are they aren’t seeing any problems, however, I am not happy with this result.



Any pointers would be gratefully received as I am at a loss to understand what’s going on here.

[saeset23]

Hello, I ran into a very similar situation this summer. By default, the gig uplink ports on the 4500's have jumbo frames enabled while the 5500's do not. Make sure to enable jumbo frames on the 5500 ports and that should put an end to any discards you're seeing.

[kjones1]

You may have hit on something there as when i checked I found that the Gigabit ports on the 4500 werent set to "undo jumboframe enable" even though i thought that this had been done. So i have disabled jumbo frames on these ports (25/26/27 & 28)and cleared down the alarms in IMC and will keep monitoring the situation to see how it goes. Ill let you know my findings. Thanks for the reply though, its good to know some people are there when you think your loosing your mind!!! LOL

[kjones1]

We'll you wont believe it, I'm still seeing the error discards on the 5500. Here's what i have done to reduce the level of discards:-



1) configure the 4500's Gigabit ports to not use jumbo frames

2) Configure the 4500's for IGMP-snooping

3) Configure the 5500's for PIM-DM



All of this has significantly reduced the amount of discards I am seeing, howver periodically i still get these errors, as seen by IMC.



N.B. For any 3Com person monitoring these channels, your documentation is shockingly poor. Details need to be more clear and specific rather than fuzzy!!!!