Cityfibre UK - ECI ONT B-FOCuS 0-1G + Mikrotik Dropouts at 2h 24m Intervals

RouterOS General
1

Good Afternoon,

This is to ask if anyone else has had issue with Cityfibre / ECI ONT’s and Mikrotik.

We have a CCR1009-7G-1C-1S+ in conjuction with a ECI ONT B-FOCuS 0-1G and every 2h 24m the interface drops. When looking at the ONT you see all the lights go off like the ONT was rebooted. We see this logged on the routerboard as the interface dropping for 60 seconds (the time it takes for the ONT to reboot).

Video here https://youtu.be/WQHpJHIlXFc

We have seen this happening with the latest v6 and previous versions of v6

Things we have tried and their results:

  • City fibre replace ONT - No Change dropout at 2h 24m


  • ONT plugged into a dumb switch - ONT stays on


  • ONT plugged into router but no fibre connected to ONT - ONT reboots


  • Put a switch between ONT and Router - interval seemed to change from 2h 24m to 4h 15m


  • Used the same model router with config restored by export - No change interval at 2h 24m

I know of another customer in the same area with Mikrotik and same ONT with Cityfibre also having drop outs.

Any suggestions?

2

That’s an odd one. As the problem still occurs with the fibre disconnected it must be something sent from the Mikrotik triggering the reboot of the ONT - possibly some memory exhaustion caused by discovery packets if you have that enabled on the Mikrotik.

3

discovery settings are all on, i’ll turn off and see if it makes a difference.

thanks

4

so far looks like it is discovery packets causing the ONT to reboot, thanks TDW.

We are testing now to narrow down if its CDP, LLDP or MNDP.

5

It may just not handle unexpected broadcast/multicast packets, which would be utterly abysmal coding in the ONT firmware.

I believe Mikrotiks only listen for CDP, they listen for and send LLDP and MNDP. As adding the switch between the ONT and Mikrotik changed the interval before reboot it is likely that both LLDP and MNDP will have this effect. (Even dumb switches, as long as they comply with 802.1D, do not forward packets to destination addresses 01:80:C2:00:00:00 - 01:80:C2:00:00:0F and LLDP uses one of these).

If you wish to use discovery internally create an interface list e.g. /interface list add name=discover and use this for discovery /ip neighbor discovery-settings set discover-interface-list=discover then add interfaces of interest to the list.

6

After testing it would seem that it’s MNDP packets causing the issue with the ECI ONT.

Tested having CDP, LLDP and MNDP on in isolation and ONT remains up with CDP or LLDP.

However enable MNDP and the ONT reboots after 2h 24m.

Creating an interface list with the relevant interfaces baring the one connected to the ONT and enabling discovery settings for that list is the solution.

Guessing Cityfibre pushed an update to the ONT that doesn’t like MNDP packets where previously it was okay. (this is based on when updates where applied to our routerboard vs when the problem started)

Thank You TDW, this problem had been going on for nearly 8 weeks.

7

Most odd. MNDP packets are not really anything special, just UDP packets broadcast to 255.255.255.255 on IPv4 and multicast to the ‘all routers’ address on IPv6.

It needs fixing by ECI, but I suspect trying to relay the information through the usual support channels will be almost impossible.