Switch - blocking access to radios

[eomedia]

I've had a weird issue lately that I'm not quite sure the cause of or where to start investigating.

I've got a ws-12-250-dc switch with a mix of UBNT radios - and from time to time the radios in my monitoring software will report as offline. I try to access the radio via web and radio is non-responsive, however, if I log into the radio on the other side of the link it shows as up and connected and traffic is flowing so the radio reported as offline is working.

Back at the site I'll log into the router and use discover and/or ping and I can see the radio and the ping comes back with 0% loss, however, if I ping from the Netonix switch(where the radio is connected) there is 100% packet loss.

A reboot of the switch will clear the issue and everything works correctly.

In the last week I also replaced the switch with a new one and the same issue is happening so it's not specific to a specific switch.

I'm not sure what information would be helpful to even investigate the issue so I'm posting this here as an initial query and we'll go from there.

[mike99]

Look at the STP tab to be sure port is still in forwarding mode. If STP is enable on radios, you should disable it. STP should also not be enable on ports facing CLI so powering AP.

Have you tryed rebooting the radio instead of the switch ? Else enabling and disabling the port on the switch ? Is the port still enable and state at 100 Mb/s or gigabit full duplex ?

Anything in the log ?

[eomedia]

STP mode is in state "forwarding" on affected ports on switch

STP is not enabled on the radios

Although I can PING the radio from the router, I cannot PING it from the switch, from within the network, access web GUI or connect via SSH to the radio - so unable to remotely reboot the radios

Ports show link correct (i.e. 100M-F on a RocketM2 link) and traffic is flowing through the link - just cannot access the radio connected to the switch.

Nothing in the log that I could see.

Thanks

[mike99]

If I understand right, the schema is like this

Router <-> Swtich <-> Radio ?

Are you sure that router ping the radio through the swich ? Have you done a traceroute ?

Look at the mac-table of the switch if radio of the mac-address is still their.

[eomedia]

Correct in the order - Router --> Switch --> Radio

In the MAC table of the switch the MAC of the radio shows, then the port, VLAN ID (1) and manufacturer and IP.

If I traceroute from the router to the radio when everything is working I get only 1 hop and it's direct to the radio - no idea if it goes through switch or not. The Routing tab shows the destination subnet as eth0 connected.

[mike99]

You should test those eh en the problem happen.

[eomedia]

I absolutely will test those and report back if/when it happens again. Thanks

[ClintFri]

I absolutely will test those and report back if/when it happens again. Thanks

Yeah please report back. That seems very odd. I feel like it shouldn't happen, I'm no expert though.

[Ludvik]

I have similar experience. Access to UBNT radio connected to netonix not working ... but forwarding to client through same radio is OK.

On router missing ARP record of this UBNT radio. I don't understand why. Maybe it's causing a multicast ban on the port. But I can not simulate it reliably enough. This happens only in the production network, not in the testing lab.
I solved it with a regular ping every minute from the router. ARP not timeouted and everything is ok ...

Version 1.4.6 and 1.4.7

[eomedia]

It happened again this morning so I was able to get a little more data.


VRRP Router SETUP
R1 - ERPoE5 (eth0) --> WS-12-50-DC - Port 11
R2 - ERPoE5 (eth0) --> WS-12-50-DC - Port 12

WS12-250-DC Port 4 - AF5X - 10.255.1.70 [ Remote Radio 10.255.1.73 ]


Issue: When the issue occurs the radio at 10.255.1.70 reports as offline

• You can NOT access 10.255.1.70 via web GUI - but CAN access 10.255.1.73
• 10.255.1.73 shows that the link is connected and passing traffic
• from netonix switch you can NOT PING 10.255.1.70 - but CAN ping 10.222.1.73
• netonix switch MAC table shows 10.255.1.70 and 10.255.1.73

Reboot Actions

Rebooting R1 and R2 has no effect and radio still shows as offline

Rebooting netonix switch resolves issue and radio reports as online

Potential Cause:
I am NOT a network guru so this is a bit of a guess on my side but seems relevant. When the issue is happening doing "show vrrp summary" on R1 and R2 shows that the VRRP state on both routers is backup - no master. After reboots of both routers there is no change - then after rebooting the netonix swtich one of the routers will show VRRP state master as expected and now things are working.

I'm not sure that's the "cause" of the issues or I would expect every radio connected to the switch to be unreachable if it was - which is not the case, but it seems like it might be related.