Acceptable error counts?

[sporkman]

This is a really short cable run, and I'm working with Ubiquiti to track down an issue with two of these PowerBeams losing the wireless connection just about once/hour. I have the ping watchdog on the UBNT enabled, which is what the ultimate trigger of a reboot is. I expect rebooting once an hour or so is going to slowly increment some of the switch port error counters just from the link going down/up.

Is there anything worrisome in these stats?

Reading about "rx-filtered" that's basically a catch all, and I suspect could be tagged VLAN frames that have no destination on that switch.

How about the rx jabber and rx fragments? Side-effect of port bouncing or indicating a possible cable issue? This switch has been up for 66 days, just for context on how long those counters have been accumulating data.

[Julian]

I think you're a lot closer than you realize to the root cause with your port bouncing theory. 2 tenths of a percent of total traffic over 66 days is also something I'd personally classify as a non-issue, but maybe that's just me.

[sporkman]

I think you're a lot closer than you realize to the root cause with your port bouncing theory. 2 tenths of a percent of total traffic over 66 days is also something I'd personally classify as a non-issue, but maybe that's just me.

Thanks - just wanted to clear it before digging in deeper with Ubiquiti support. The more I can rule out, the better, I count myself lucky they're actually looking at the issue outside the forums.