You're likely running into the low size of buffers on that switch. If you aren't using many ports at once you can raise the maximum size of the port queue a good bit since it just carves more out of the main shared pool on that switch. A search of the archives should find the relevant commands. You're looking for the command to set them above 100% buffer size allocation.
-Blake On Jul 19, 2013 11:50 PM, "Mikael Abrahamsson" <[email protected]> wrote: > On Fri, 19 Jul 2013, Jerry Bacon wrote: > > When the the ports facing the radios are in access mode, everything is >> good, but when they are in trunk mode (which we need), we get 2 - 3% packet >> loss at small packet sizes (high packet rates). We've tracked it down to >> output queue drops, but I don't understand why it's different for access >> vs. trunk. >> > > Please describe more in detail what your configuration is like. > > If the fluke is connected to an access port, you then vlan tag the traffic > on the trunk port, then you're adding 4 bytes per packet meaning your > access port is no longer wire speed and the loss you're seeing would be > understandable. > > Any ideas? And would QOS tuning be likely to help? >> > > QOS is a way of selecting what packets to drop. I don't see how it would > help here. > > -- > Mikael Abrahamsson email: [email protected] > ______________________________**_________________ > cisco-nsp mailing list [email protected] > https://puck.nether.net/**mailman/listinfo/cisco-nsp<https://puck.nether.net/mailman/listinfo/cisco-nsp> > archive at > http://puck.nether.net/**pipermail/cisco-nsp/<http://puck.nether.net/pipermail/cisco-nsp/> > _______________________________________________ cisco-nsp mailing list [email protected] https://puck.nether.net/mailman/listinfo/cisco-nsp archive at http://puck.nether.net/pipermail/cisco-nsp/
