GL-MT6000 (Flint 2) Multi-Wan

Your question isn’t a duplicate of my question. It’s about how to assign tethering addresses, not about using multi-wan. (ETA: You’re using failover, I’m not. I’m trying to use load balancing, which is a different application of multi-wan. There is no reason to combine the two into a single topic - Discourse, the platform that they’re using here, has no limit on the number of topics.)

What doing this does is make me (as the person who asked the original question) think that progress has been made on my question - when it hasn’t.

Please start your own topic so you can effectively track your issue and I can track mine. :slight_smile:

1 Like

Hi, wanted to check back and see what else I should be looking at to get load balancing working with this router. As noted previously, making the suggested changes doesn’t seem to have made the connection stable when configured for load balancing.

Is there a detailed configuration guide that outlines best practices for configuring this? (I realize it’s a new feature, so such a guide might not exist, but I figured it wouldn’t hurt to ask. :slight_smile: )

Sorry I didn’t notice this earlier, but it’s best used when you need to download content from multiple sites at the same time.

I’m aware of what the use case is for load balancing. That’s why I’m trying to configure it.

But also given that the WAN ports on this router are each capable of 2.5 Mbps of throughput and the LAN and WIFI interfaces aren’t capable of exceeding that, it is unlikely that a LAN-connected device is going to max out the connection speed on one port, let alone two.

I’ve been working on networks for over 30 years, and wrote documentation for a reverse proxy product that load balanced connections to web servers and accepted input from F5-style hardware load balancers. I understand what the functionality is, and I have configured it before on many different hardware and software devices. (I recognize you have no way of knowing this information without me telling you, so I’m sharing my background in hopes that moves the conversation along to a resolution, because a description of the feature isn’t helpful with my experience :slight_smile: - I know what it does.)

My question is why I can’t get it working here. Configuring load balancing on the WAN ports should have no effect on the functionality from a single client. But when I configure it on this device, it does - I have intermittent connectivity failures from my main desktop (connected directly to one of the LAN ports) when accessing sites on the Internet when load balancing is configured.

I’m looking for technical support here to determine (a) why it isn’t working, (b) what troubleshooting steps I can take, and (c) what configuration I may have missed to get it working smoothly.

I could think of some bug like the data stream will be split over two WAN connections and therefor the order of packages is broken and the TLS tunnel fails.

Yeah, I had a similar thought (not about the packet sequence): If the connection is somehow not pinned to the proper IP address so all traffic for that specific connection persists, that could explain the drop. I suppose I could put tcpdump on the router and try a capture of a failed connection. From a browser perspective, I’d expect to see a “network changed” message rather than just a failure, though.

Thank you for the background.Here are some of my thoughts:

(a). The load balancing egress route is inconsistent with the previous route, causing the website to fail to load. Try refreshing it to see if it can be restored.

(b) You can use tcpdump to capture packets on interfaces that participate in load balancing. You can see that the packets that are successfully accessed are not captured on the same interface when the access fails.
(c) Nothing else needs to be configured. You can use the ip rs command to check whether the load balancing is effective.

The picture is an example: the red box indicates the interfaces involved in load balancing, and the numbers in the yellow box indicate the load ratio

Thanks for this info - I’ll take some time later this week (I have some time Friday when I can look at this without disrupting other users on the network) and see what I can find out.

Am I understanding correctly that the load balancing software doesn’t pin the session to a particular interface? If that’s the case, that seems like it would really limit the usability of this functionality. It almost might be better for me to just use a VLAN configuration to automatically route a specific device to a specific port and manually isolate the traffic that way.