Hi
Hoping someone can help with this, my RB3011 is showing that there is a loop and there isn’t one.
To elaborate,
Eth1 goes to a CRS125 and it has nothing in the logs suggesting any kind of routing loop
Eth2 goes to an RB2011 and it too has nothing in the logs suggesting some kid of loop
Eth3 goes to a Vonage VOIP box which has no way of creating a loop (and the problem existed before the box)
Eth10 is running PoE up to a roof mounted Ubiquiti CPE and I run a pppoe client on Eth10
Whenever it’s happening, run a sniffer on the interface and capture it to a file (e.g. broadcasts.cap) - let the sniffer run for a little while and then stop it, and download the file and open it in wireshark. Note the source of the broadcasts and track it down that way.
I have been pulling my hair out with this one. So far I’m yet to try the above but I just had a gut feeling it may be the RB3011 itself at fault.
I did not have any issues up until a few days ago and no dramatic changes in my network. I have been enabling an disabling ports on the adjoining switches to try and rule out where the source is coming from, every time the problem seemed to follow no matter what was disabled.
***With just my Vonage and Ubuntu PC connected to the RB I was seeing over 25Gb going over ether7 which was not in use!
I am currently using a hAP lite in place of the RB3011 and so far 7 minutes of error free logs with my network “as it was” when the RB started to have a tantrum.
We used to have an issue sometimes where after lightning storms (very very common in my area), sometimes a wireless station would go nuts and create a loop - so as soon as this “mad station” would join the wlan, the wlan would get swamped by broadcasts and looped frames, causing the switch the AP was connected to to errdisable the port. We had to put a MAC acl in place and bring up stations one-by-one until the problem hit again…
This happened even when the rogue station had no link on its ethernet side (tech physically disconnected the cable)…
I guess this is all to say that it really seems to me that something has gone haywire and is apparently really causing a loop on the network. It could be the RB3011 itself, or there could be something in your hAP Lite’s configuration that prevents the snowball from rolling down the hill, but is missing in your 3011 configuration…
I didn’t quite follow your meaning on this:
So the “tantrum” happened even when you put the hAP lite back in place, or do you mean “when the (3011)RB started to have a tantrum?”
Sorry it was quite badly worded. I’ve swapped out the RB3011 for a hAP lite that I had knocking around the office and the hAP is playing nicely with no errors and no routing loop messages. At that point I’d been rifling through my patch cabling moving everything around trying to resolve a possible hardware loop.
My LAN is entirely back to where it should be now as it was when the RB3011 first started complaining of a loop only now I have the hAP in it’s place and that is working fine.
This draws me to 3 things currently; Faulty RB3011, Bad version of rOS or bad configuration of rOS by me.
I’m going to factory default the RB over the weekend and see if I can configure it back up in a basic manner to make it work again and if not the RB will be making it’s way back to MikroTik land.
I was hoping the same but I’ve just finished reconfiguring it, did it from fresh factory default then reset config with no default and it started throwing errors and mad amounts of data over unused ports.
I think she will be going back. Bit of a knock as I generally shout MikroTik from the rooftops, I just hope the distributor can follow up with some good service.
Have tried it with 6.34, 6.34.1 and the 6.35rc8 and got the same thing.
I saw in previous bug fixes a false positive message had been removed but that doesn’t explain the 25Gb traffic across an ethernet interface that isn’t being used.
Also I did have the ports in a “dumb” switch off ether1 as master but when problems started tried moving them into a software bridge to see if that helped and still got them same thing on them individually.
I’m having the exact same issue with the RouterBOARD 3011UiAS on firmware 3.27.
Im’ very new to routerOS and I’ve implemented the following config:
Config:
ether 1: connected with pppoe interface
ether 2: master port
ether 3: slave to 2
ether 4: slave to 2
ether 5: slave to 2
ether 6: master port
ether 7 : slave to 6
ether 8 : slave to 6
ether 9 : slave to 6
ether 10: slave to 6
I’m getting a dynamic IP via PPPoE connected on the ether 1 port. I wanted the rest of the ports to function as a switch an after some reading on the wiki I came up with the following config.
to make switch port you need to make 1 master ports and set the other ports to slave on the master port. Because the 3011UiAS has two switch chips I needed to config this with the two ports : eth2 and eth 6.
\
To connect the both switch groups (eth2-5 and eth6-10) I’ve made a bridge interface and added the both master ports to the bridge (eth 2 and eth 6).
Gateway IP assigned to the bridge interface, DHCP, Natting etc..
Is there any flaw in my configuration implementation that could cause the errors ?
-I’ve made the bridge MAC static
-checked cabling
-port status on bridge ports are designated, default settings rtsp.
-disabled auto negotiation and set the speed and duplex mode static
You shouldn’t have any loops due to this configuration. But the giant flood of traffic tells me there really is something out there repeating broadcasts. The logs are warning about excessive broadcast, and your observation of phantom bandwidth consumption supports this.
You could wait for the next broadcast storm and then start disabling ports one at a time until it goes away, and that would tell you which interface is involved in the flood.
Sounds like we are getting same problem with s same firmware. Which ros are you running?
Shame a mikrotik person can’t comment on this, it’s a little convenient 2 independent networks with no issues get this message with the same router which is a relatively new release.
I’m running:
Board Name RB3011UiAS
Version 6.34.1 (stable)
As suggested i’m eliminating ports by adding my clients one at the time and running data through the ports.
Config:
ether 1: connected with pppoe interface
ether 2: master port
ether 3: slave to 2 → 1 connected desktop client(w10). No problem with collisions/routing loop errors
ether 4: slave to 2 ->Synology nas, no bonding/802.3ad just 1 connected utp on the port. Gigabit with auto negotiation.
ether 5: slave to 2
ether 6: master port
ether 7 : slave to 6
ether 8 : slave to 6 → raspberry pi
ether 9 : slave to 6
ether 10: slave to 6
After adding a third client in the switch group the router started signalling router loops and collisions. Collisions were also reported on disconnected ports.
I disconnected the client on port eth8 (in the second bridge group) and rebooted the router. I did this because Isuspected a problem with the bridging between the two switch groups. Collisions were reported again on the master port eth2 but none were reported on the interface where both clients were connected (eth 3 and eth4).
I disabled the bridge interface that connected both the master ports (eth2 <->eth6) and rebooted the router. The only clients that are connected are on eth 3 and 4 (and eth1 my ISP modem for the PPPoE)
Config:
ether 1: connected with pppoe interface
ether 2: master port
ether 3: slave to 2 → 1 connected desktop client(w10). No problem with collisions/routing loop errors
ether 4: slave to 2 ->Synology nas, no bonding/802.3ad but 1 connected utp on the port. Gigabit with auto negotiation.
ether 5: slave to 2
ether 6: master port
ether 7 : slave to 6
ether 8 : slave to 6
ether 9 : slave to 6
ether 10: slave to 6
NO collisions/loops or duplex errors are reported thus far.
Question: It seems to me that the problem is with the usage of a bridge interface, I’m new to Mikrotik so I’m not familiar with this kind of interface. Is it possible to replace the bridge interface (which is pure software based) with a UTP cable between the two master ports and in doing this connecting the both switch groups ? I’ve read that the master ports cannot be used though to transfer data, do I then have to connect ports from both switch groups e.g. eth 5 ↔ eth10)?
Update: I’ve connected my MikroTik RB260GS switch that has a couple of client (AP, media server etc) on port 5. Still no collisions/loops/errors reported.
After an entire night I still got no loop/collisions or duplex mismatch errors. It seemed the problem was with the use of a bridging interface.
Is there anyone who could explain this to me ? I thought I understood the use of a bridiging interface in connecting two different switch groups but it seems like I was wrong ? (or is it a bug in this model/firmware?)
to make switch port you need to make 1 master ports and set the other ports to slave on the master port. Because the 3011UiAS has two switch chips I needed to config this with the two ports : eth2 and eth 6.
To connect the both switch groups (eth2-5 and eth6-10) I’ve made a bridge interface and added the both master ports to the bridge (eth 2 and eth 6).
Gateway IP assigned to the bridge interface, DHCP, Natting etc..
No not wrong at all. The bridge interface is designed to do as you are intending. I ran the same setup on my RB2011 with no problems at all, only snce moving to RB3011 is this occuring.
This would also explain why when I removed all ports from the master-slave setup onto a software bridge the problem seemingly got worse. I don’t think not using a software bridge though is an ideal solution and MT need to come up with some answers to this.
I have mailed this to the support addres of Mikrotik. In the meanwhile, if someone has any further troubleshooting ideas I’d be happy to try them out. I still have the following questions though:
Is it possible to replace the bridge interface (which is pure software based) with a UTP cable between the two master ports and in doing this connecting the both switch groups ? I’ve read that the master ports cannot be used though to transfer data, do I then have to connect ports from both switch groups e.g. eth 5 ↔ eth10)?
If this is not an option, is it only possible to use the ports from the second switch group (eth6-10) as router ports?
You can definitely use the master ports. My CRS125 which resides below the RB3011 has master port of ether1 and the link from the RB goes into that.
You could with no problem run a patch between the 2 switch chips as a short term fix but that just seems so counter productive when the software is malfunctioning causing this. Why bother with a 10 port router when you have to use 2 to join the thing together?
I would have preferred all ports on same switch chip either way and have always seen this as a weakness in the RB’s.
Thanks for the reply, I’ll try the patching tonight and see if encounter any problems.
I agree, this solution isn’t quite that elegant and I’m glad I bought the router with no expectations whatsoever. I just wanted to test the hardware specs, get to know the Mikrotik OS and replace some equipment by one device.
At the moment it is replacing a Cisco router and a Cisco switch in my home enviroment but with this problem I’m missing some ports and will have to add my switch again.
