How to force XT8 mesh into particular topography config

[DraconianCurve]

Hi,

I've searched for an answer to this, but none was forthcoming.

I have an unusual layout at home in terms of where the rooms are relative to where the internet comes in. To fix this, our main router, positioned where the fibre comes in, then connects wirelessly to a node positioned centrally, which then in turn connects to two further nodes positioned at various 'remote' places... (It's not a particularly large house - just an unusual layout...)

At least that's how it's supposed to be. But as much as I try, including setting each node's 'Preferred WiFi Uplink AP', I can't get this setup to stick... It keeps reverting to all three nodes connecting to the main router with weaker connections than they would have if they just stuck to how I want them.

Is there a way to fix this so it doesn't keep happening? Any settings which might, at least, improve it a little?

 

[pvanosta]

what happens when you ask the system to optimize itself? Same topography result?

 

[raven-au]

Hi,

I've searched for an answer to this, but none was forthcoming.

I have an unusual layout at home in terms of where the rooms are relative to where the internet comes in. To fix this, our main router, positioned where the fibre comes in, then connects wirelessly to a node positioned centrally, which then in turn connects to two further nodes positioned at various 'remote' places... (It's not a particularly large house - just an unusual layout...)

At least that's how it's supposed to be. But as much as I try, including setting each node's 'Preferred WiFi Uplink AP', I can't get this setup to stick... It keeps reverting to all three nodes connecting to the main router with weaker connections than they would have if they just stuck to how I want them.

Is there a way to fix this so it doesn't keep happening? Any settings which might, at least, improve it a little?

None that I'm aware of.

Are you sure it's actually better to have it the way you think it should be.

Using the "Preferable Uplink AP" has to bias the selection so maybe the difference is great enough that it's not being used.
For example, for the uplink to be better going via a nearby node it would need to be at least twice the sync rate to give similar performance because the wireless back haul would then be shared by the node plus the other node going via it.
So maybe the mesh nodes aren't being so silly after all.

 

[DraconianCurve]

what happens when you ask the system to optimize itself? Same topography result?

It varies, sometimes it aligns with how I want it, and othertimes it doesn't...

 

[DraconianCurve]

None that I'm aware of.

Are you sure it's actually better to have it the way you think it should be.

Using the "Preferable Uplink AP" has to bias the selection so maybe the difference is great enough that it's not being used.
For example, for the uplink to be better going via a nearby node it would need to be at least twice the sync rate to give similar performance because the wireless back haul would then be shared by the node plus the other node going via it.
So maybe the mesh nodes aren't being so silly after all.

That's an interesting way to think of it, and it makes sense.

All I can say is that the performance I see from devices connected to various nodes is better when the nodes themselves link up the way I want. It's difficult to measure it objectively, though, given what you've said. The signal strength between nodes when connected as I wish is -50dbms or better. This goes down to -65dbms or lower when connected as it often does, with then all linking direct to the main router.

Perhaps the answer, as it so often is with these things, is I should just stop worrying about it and let them do their thing...

 

[leerees]

Select the unlink biases you'd prefer. Once complete, goto AI mesh and click optimise. This also fixes the issue of nodes dropping off the network.

 

[DraconianCurve]

Select the unlink biases you'd prefer. Once complete, goto AI mesh and click optimise. This also fixes the issue of nodes dropping off the network.

That's how it should work, but it doesn't appear to for me. All the preferred uplink settings are as I want them, but the optimise function seems to rarely pay any attention to it. I just did it again, just to make sure I'm making it up, and I'm not...

 

[leerees]

It worked for my AX92u's. I didn't even need to set any bias. As soon as I clicked optimise they rearranged into a star topology.

I'm interested in getting XT8's for a new network I'm building. What's the latency like compare to wired? The AX92u's add 1-2ms on WiFi. I'm hoping the XT8's offer the same performance.

 

[DraconianCurve]

It worked for my AX92u's. I didn't even need to set any bias. As soon as I clicked optimise they rearranged into a star topology.

I'm interested in getting XT8's for a new network I'm building. What's the latency like compare to wired? The AX92u's add 1-2ms on WiFi. I'm hoping the XT8's offer the same performance.

Is your main router at the centre of your star arrangement? If so, perhaps that's the difference to my setup and why it works for you.

Would be happy to help with your question on latency. Do you know if this is measured anywhere within the router's GUI? I don't recall seeing it anywhere...

 

[leerees]

Is your main router at the centre of your star arrangement? If so, perhaps that's the difference to my setup and why it works for you.

Would be happy to help with your question on latency. Do you know if this is measured anywhere within the router's GUI? I don't recall seeing it anywhere...

The router is physically in the middle. The satellies are all placed evenly around the router to form a flat square shape (it's a block of apartments). We have a NAS server connected to the main CAP and 2 additional servers which mine crypto currency connected to a node. There's been no issues with stale hashes.

I tested the latency by pinging Google. AX adds 1ms and AC adds 2ms. I didn't test 2.4ghz but with the AX92u's I beleive it's only N.

The AX92u 2.4ghz will give 100mb/s so we have the IOT devices and older streaming TV's on that.

 

[raven-au]

That's an interesting way to think of it, and it makes sense.

All I can say is that the performance I see from devices connected to various nodes is better when the nodes themselves link up the way I want. It's difficult to measure it objectively, though, given what you've said. The signal strength between nodes when connected as I wish is -50dbms or better. This goes down to -65dbms or lower when connected as it often does, with then all linking direct to the main router.

Perhaps the answer, as it so often is with these things, is I should just stop worrying about it and let them do their thing...

Mmm ... that -65dbm is quite low IMHO, the selection algorithm is probably set to be a bit too pessimistic but ...
It will be interesting to see if you find a way to influence the selection in favor of what actually performs better for you.
IIRC the last time I had connection drop outs I had to change the device location (less obstacles, closer), it was connecting at somewhere between -65 and -70 dbm.

 

[Stepalex]

Same issue for me, XT8 nodes ignore my Preferred WiFi Uplink AP setting and choose a vastly suboptimal topography, defeats the whole purpose of the mesh(

 

[wsalopek]

Old thread - but, maybe you can orient the router antennas to force a higher or lower signal strength between the routers themselves? Like set the dipoles antennas 90 degrees to each other - certain router antennas (ALL of the antennas on a given router) placed vertically, and other router antennas placed horizontally).

The router antennas at 90 degrees to each will transmit less signal strength than the antennas parallel to each other.

Of course this MAY negatively (or positively) affect the signal strength some devices will receive, but hopefully for them, it'll be 6-or-a-half-dozen-of-the-other.

P.S. The Netgear Orbis allow a person to force a given topology.