hEX PoE with router OS 6.40.4, Fiber uplink to the main switch, PoE powered phone on an Ethernet port, device should act as a switch, but:
I can ping the admin IP of the hEX and login via the SFP port, so the fiber link is OK. But pinging the phone results in a packet loss rate of 70-90%. Consequently the IP phone does not register.
Using an Ethernet port for the uplink, everything works as expected.
How can I configure the hEX to act as a switch, switching all 6 ports including SFP? I switched the hEX to bridge mode, but left everything else to their defaults.
Strange is the packet loss. I would expect it working or not, but not only for some packets.
Any clue? Thanks!
On the website, for every product you can find a link to a “block diagram”.
For the hEX PoE, you can find it on its product page: https://mikrotik.com/product/RB960PGS
where the block diagram link points here: https://i.mt.lv/routerboard/files/RB960PGS-161220141841.png
Now that you have looked at that block diagram, what is your expectation about switching the SFP to the ether ports?
This of course does not explain the packet loss unless the router is under full load from other traffic, which you do not mention.
So you need to research where the packets are dropped. On the link, in the router, somewhere else?
Thanks for the hint, @pe1chl ! Maybe wrong wording, “wirespeed switching” between SFP and Ethernet ports is obviously not possible. But shouldn’t be a bridge between ALL ports created in “bridge mode”? Or is there any special configuration necessary?
There shouldn’t be high load on this device as there is just an IP phone and an up-link link connected.
Yes it should work with a bridge between SFP and ether1 (and ether2-5 as slave to ether1).
I have no idea why you have packet loss.
I suspect there are problems for hEX poe running 6.40.4, I’ve recently downgraded two units to bugfix 6.39.3 because poe autodetection on ether2/3 showed false shortcircuits (4/5 weren’t affected). I would recommend to try bugfix 6.39.3 and report back
Thanks for this, bajodel! Actually I was sure that I had blown up the PoE circuit of Ether2. I got the hEX PoE just a few days ago and using it in an experimental way, trying to master the most important functions. I thought I might have accidentally plugged an hEX PoE port to another switch’s PoE port. But those short circuit notices disappeared after downgrading to 6.39.3. All PoE Ether ports OK now.
Anyway, still have to sort out how to use the SFP as bridged port.
Some more details:
- SSH into the hEX console does work using the SFP (fiber) or any of the Etherports
- Pinging from the heX console to my PoE powered devices works OK, as long as I SSH over another Etherport
- Pinging from the heX console to my PoE powered devices does not work OK (90-100% packet loss), as soon as I SSH over the SFP
It seems something happens as soon as I connect the fiber.
Edit: Log message: sfp1: bridge port received packet with own address as source address (64:d1:54:bb:49:c1), probably loop, that’s ether-2 (master)
This is the default bridge config. Only DHCP deactived and fixed admin IP given.
Any clue?
Try setting the Spanning Tree protocol of the bridge to “none”.
There is some incompatibility with certain other manufacturers.
OMG. Thanks everyone! It is so embarrassing.
I forgot that I dedicated the port on my main switch used for the hEX fiber uplink for … port mirroring. ![]()
The good thing is that in the course of this discussion I got the hint about the PoE short circuit being actually a SW bug and learned a few other things.
I think MikroTik produces some quite decent devices. Of course users should know what they are doing…
The bridge works now as expected.