Lots of Errors with Netonix and new 11x

[mhoppes]

I'm not sure what to make of this. As far as I can tell there are no physical issues on the Ethernet medium. And I've never seen this before.

This is a PtP link:

EdgeRouter--->Netonix--->11x~~~11x--->Netonix--->EdgeRouter

There is a /29 between the edgerouters. MTU on the switch is 1528, on the edgerouter is 1500. Why am I seeing oversized and jabber errors, to begin with -- with such huge quantity?

These errors are counting up hundreds per minute.

[mhoppes]

Changing the MTU on the switch from 1528 to 2000 caused JABBER and FILTERED RX errors to stop counting. CRC errors are still counting up slowly.

Why would a larger MTU, when the router, which is sending packets to the radio is set to 1500, stop the JABBER and FILTERED RX errors?

[mike99]

Seem like the AF11 add some kind of encapsulation to the data. I wouldn't be able to say why since I don't know how it work under the hood but paquet capture and check this out with wireshark could give the answer.

[jefa]

The edgerouter probably reports the IP MTU, while the Netonix switch uses the mtu of the entire ethernet frame, so you have to add the size of the ethernet header/encapsulation. If you add vlan-tagging/QinQ, you will have to increase the mtu on the switch.

[mhoppes]

There's literally no VLAN going on on this connection though.

[lligetfa]

CRC errors are still counting up slowly...

I've only ever seen CRC errors on a layer1 issue.

[sirhc]

CRC are "usually" cable related.

If you move link to another port and the errors are still there it is most likely cable.

If Errors stop either the connector was loose or the switch port Ethernet Transformer or PHY may be at fault.

You could also have a bad port in the radio.

When AF24 first came out I had several units that I had to RMA back to UBNT.

We determined it was the radio port after first changing cable and ends with no love
Then a different Ethernet port in Cisco router (Before Netonix existed)
Finally swapped out AF24 with spare problem solved so it was the port in the radio. (IT HAPPENS)
Today I would simply ask Julian to replace the Transformer and possibly PHY as we have been repairing UBNT / MIMOSA / MikroTik equipment for customers for going on a year now.

[mhoppes]

I've swapped:

CABLES
ENDS
RADIOS
SWITCHES

When plugged into another brand of switch these errors do not happen. This leads me to believe there is a problem with the Netonix switch and the way it is interacting with the airFiber11 radio.

[sirhc]

There are other users who are using our switches with AFX11 with no issues?

Considering the AFX11 uses the almost the exact same board design as other AFX radios, exact same PHY and Transformer?

Does not make sense?

[itwnz]

When plugged into another brand of switch these errors do not happen. This leads me to believe there is a problem with the Netonix switch and the way it is interacting with the airFiber11 radio.

Here's 2 examples... no real problem with errors but I must say the stats on all other equipment connected to these switches are a lot cleaner.

AF11X and WS-12-250-DC - 122 days uptime:
Rx Drops 50460
Rx CRC 2
Rx Undersize 0
Rx Oversize 22
Rx Fragments 1
Rx Jabber 66
Rx Filtered 12988608


AF11X and WS-12-250-DC - 36 days uptime:
Rx Drops 2736
Rx CRC 7
Rx Undersize 0
Rx Oversize 634
Rx Fragments 0
Rx Jabber 1993
Rx Filtered 1669