Hello,

Anyone else here experiencing a total loss of comms when sfp1 port on hEX S is disabled in RouterOS?
I have tried with 6.42.9 and 6.43.4, updated routerboot, however it is still the same and happening on two devices I have tried so far.
Rest of the ethernet interfaces can be bridged or out of the bridge but as soon as I will disable sfp1 I will lose access to RB via any ethernet port (they still will be linked up, just won't be passing data to CPU it seems).
Only way to recover from this state is to default the config via the button.

Thank you.

Is there a possibility, that you have a default config address assigned to sfp interface instead of the default bridge?

Hello,

Nope.
IPs were assigned either to bridge iface or to ethernet ifaces.
MAC telnet was enabled as well but I was not able to see the router in Winbox neighbors nor was able to login to the router over L2.
And there were no FW rules.
Anyways, I was about to call it a day and it recovered.
It looks like after disabling sfp it takes (in this case) around three minutes to recover:


Device changed by admin was me disabling SFP. Was kicked out right away.
Was not able to ping bridge iface (only bridged ether2 through 4), was not able to see the router in Winbox and MAC Winbox was timing out.
Then three minutes later logs show port flapped.
And now I can log in and sfp1 is showing as disabled.

I guess it has to do something with that data lanes XOR logic - either SFP or one more lane to the switch chip.
Anyways, problem solved it seems.

Thanks.

Is there a possibility, that you have a default config address assigned to sfp interface instead of the default bridge?

I guess it has to do something with that data lanes XOR logic - either SFP or one more lane to the switch chip.

You are right - looks like most possible cause.
Despite you say the problem is solved, I think this behaviour should be described somewhere (on a wiki? in quick start guide?) in a form of warning.

Can confirm and I've reported this already under #2018091622001254 - they're fixing it in future versions. Do note that not all ports are affected - check block diagram.

I guess it has to do something with that data lanes XOR logic - either SFP or one more lane to the switch chip.

You are right - looks like most possible cause.
Despite you say the problem is solved, I think this behaviour should be described somewhere (on a wiki? in quick start guide?) in a form of warning.

Eh, I spoke too soon.
I have set up a switch group (ether2 through 5).
Once SFP slot is disabled, ports 2 and 4 are not communicating and will not recover. Ports 3 and 5 are okay, not affected by disabling SFP.
After you powercycle the router all is okay, even on ports 2 and 4.
You can enable SFP without any problem, but as soon as you disable it again, you will loose comms on 2 and 4 again.
Disconnecting and reconnecting ethernet doesn't work, what only works is powercycle or ports 3 and 5.
Port 1 doesn't seems to be affected in this configuration (separate port).
If the ports are not in a switch group they seem to recover after a while.

Can confirm and I've reported this already under #2018091622001254 - they're fixing it in future versions. Do note that not all ports are affected - check block diagram.

Hello, thanks, and do you maybe know if future software or hardware (revision) versions?
I hope software, cause I was thinking about getting these units back (as there was this SFP linking issue before which apparently was fixed by new HW revision) and ordering HEX PoE instead as we used to.
Can someone with Mikrotik confirm it can be fixed in software and what is the ETA?