Enabling link aggregation with Ubiquiti Edgeswitch

[preacher]

I have an XT9 acting as my mesh router with two XT6 nodes hardwired. It all feeds through a Ubiquiti Edgeswitch lite 24 port switch.

I'm trying to enable link aggregation with my XT9 and its not working. On the switch interface I just click the two ports, click aggregate and confirm a LAG. It then shows as linked.

On my router I am enabling it through bonding/link aggregation and using port 2&3 on the XT9, but when I do it just crashes the router into a loop and the light goes red. When it does I lose the wifi too, it shows up, wont connect, dissapears, then shows up again and so on.

I read a similar problem that just said to use port 1&2 on the XT9, but that didnt work either.

I have to unplug one of the lan cables from the back of the XT9, reboot it and disable the aggregation on the switch be fore the router works again.

Can anyone help?

 

[Ripshod]

The switch needs to support LACP (802.3ad) and have that setting enabled. I'm not familiar with that particular switch but there may be a LACP/static switch hidden in the settings somewhere.

 

[Tech9]

Can anyone help?

Even if link aggregation to your switch worked - AiMesh doesn't support backhaul link aggregation and your nodes will stop working. What you are trying to do has no practical benefits.

 

[preacher]

I'm trying to create link aggregation to the switch to double the bandwidth as I have a lot of the house hardwired, not specifically for the nodes. Is it because the nodes are hardwired its not working?

 

[Tech9]

The bandwidth to any single connection will be still limited to Gigabit, any connected to Gigabit port wired device will be obviously limited to Gigabit, your XT9 is not really 2Gbps capable WAN-LAN hardware and wired AiMesh doesn't support link aggregation on the backhaul.

 

[Boatman]

Sorry if I’m seeming argumentative, I’m just trying to understand. The idea was I would have double the bandwidth to share with all the hardwired devices. The kids have tvs in their bedrooms and consoles, there’s two nas drives (one backing the other up), computers, and a chunk of smart home stuff all hardwired. So I thought it would be useful.

If I made the nodes wireless backhaul would that fix the issue?

 

[Tech9]

Wireless backhaul will be worse compared to Gigabit Ethernet.

 

[glens]

Running a pair of (hardwired) XT8s (which suffice quite well otherwise) but wanting to increase internal network bandwidth, I picked up a GT-AX6000 with its LAN-side 2.5Gb port.

Log in (ssh) to the XT9 and start an iperf3 daemon. If it's not native to your firmware one compiled for the architecture can be readily found (pretty sure from the source of iperf3 code itself). Just copy it to the router's /tmp directory and run it from there. Either way, run a test from the router to itself. The XT8s can do about 1.4Gb/s that way. The GT similarly does right at 5 times that rate. Via the 2.5Gb connection between them, the XT8 can do about 1.7Gb/s /to/ the GT, slightly less reverse direction. So I have to agree with the

not really 2Gbps capable WAN-LAN hardware

assessment, except I surmise the limitation goes a little deeper than the WAN-LAN hardware itself.

 

[glens]

Assuming the major functional difference between the 8s and 9s lies in the 5-1 radio sets, I thought the above would be pertinent here.

[edit: further addition: Would like to "hear" the result of that test on the XT9.]

 

[tgl]

Log in (ssh) to the XT9 and start an iperf3 daemon. If it's not native to your firmware one compiled for the architecture can be readily found (pretty sure from the source of iperf3 code itself). Just copy it to the router's /tmp directory and run it from there. Either way, run a test from the router to itself. The XT8s can do about 1.4Gb/s that way. The GT similarly does right at 5 times that rate. Via the 2.5Gb connection between them, the XT8 can do about 1.7Gb/s /to/ the GT, slightly less reverse direction.

AFAICT, what you're describing is a test of bandwidth to/from the router's CPU. For the XT8s/XT9s, and most all gear in this price class, that's not a great measure of what the router can pass through to clients. The client-to-WAN data paths have hardware assists that don't apply to packets going to/from the CPU. So it's more useful to do tests using iperf3 on a client machine connecting to another machine on the WAN side of the router.

Having said that, the OP's problem seems to be that link aggregation doesn't work, which really doesn't surprise me. ASUS can barely manage to keep mainstream use-cases working on these boxes, never mind off-the-beaten-path features.

 

[glens]

Just looked at asus.com. They're hawking LAN link aggregation for that model. No manual available, and a general topical FAQ.

Ports 2 and 3 according to the captured graphic.