What's new

AiMesh Observation on "main" vs. "node" (AX vs. AC)

  • SNBForums Code of Conduct

    SNBForums is a community for everyone, no matter what their level of experience.

    Please be tolerant and patient of others, especially newcomers. We are all here to share and learn!

    The rules are simple: Be patient, be nice, be helpful or be gone!

lindros2

Regular Contributor
I've now tested AiMesh - and hacked together repeaters - with Asus for years. Yes, I use wireless backhaul, and I know the limitations.

Over the weekend (stupid me) I decided to make another attempt at using a GT-AX11000 as the "main", and GT-AC5300 as the node.
Yet again, I failed. Performance went from bad (~200mbps) to awful (~25mbps) with clients attached to the node. (NOTE - AX clients attached to the main are at or above 600mbps, which is crazy)

In REVERSE (which is what I've normally had with both RT-AC5300 -and- GT-AC5300), I get 300-400mbps (or more) both wireless and wired into the satellite node.

So... I'm wondering - is using an AX router (or AXE) a bad idea for the main, because it has crap backhaul?
This seems bizarre to me - why wouldn't you want the main unit to be most powerful? My conclusion is "wireless backhaul on new devices is terrible"
 
Yes, you do want the most powerful router as the main.

However, AX routers can't communicate at their highest speeds with AC routers, and worse, may have other bugs/limitations.

That is why you should at least match AX to AX class if you can and using a wired backhaul is even more advantageous.

Another issue today is that Asus' AX class routers are (or will soon be) on 388.xx level firmware, while all AC class routers will be on 386.xx until they are EOL (they will get updates on 386.xx level code base for a while, or longer).

What firmware are the two routers on, specifically?

What steps have you taken to ensure a clean network install when you're testing?

Are you using any amtm (or other) scripts?

Is your node too far from the main router?

Is there no possibility of a wired backhaul in your space?

AiMesh Ideal Placement

Control Channel Setup 2021

Control Channel Setup (more)

The above links may give you more hints for a more stable wireless backhaul solution with your existing routers (taken from the link below).

Almost all L&LD Links
 
What firmware are the two routers on, specifically?

What steps have you taken to ensure a clean network install when you're testing?

Are you using any amtm (or other) scripts?

Is your node too far from the main router?

Is there no possibility of a wired backhaul in your space?

AiMesh Ideal Placement

Control Channel Setup 2021

Control Channel Setup (more)

The above links may give you more hints for a more stable wireless backhaul solution with your existing routers (taken from the link below).

Almost all L&LD Links
1. Production - latest - Asus (NOT Merlin) firmware.
GT-AC5300 = Current Version : 3.0.0.4.386_48377-g3e428e2
GT-AX11000 = Current Version : 3.0.0.4.386_49599-g8352df7

2. "clean network" - not restore from backup; started clean. Performed full reset and wiped anything (I do not use scripts nor do I ever ssh in).
I also removed all devices - i changed the SSID's slightly so zero machines auto-reconnect. This didn't matter - changing back to normal still yields 2x-10x performance with GT-AC5300 (or RT-AC5300) as main.

3. no script.

4. "too far from main" - this doesn't matter - the inverse (using AC routers) doesn't see the massive performance hit.
NOTE - if the routers are too CLOSE together (oddly), it uses the 2.4Ghz channel for backhaul, shows "poor" connection, and basically fails. Which kind of makes sense.

5. ZERO chance of wired backhaul. I got a quote for $2700 to wire it - which is why I use wireless in the first place.
(I'll wire it when I bust apart my house in 2030)

6. None of the "ideal placement" stuff matters - I'll say it again - if it's REVERSED (all of these are tri-band routers), using GT/RT-AC5300 as main, it works great, but if I use GT-AX11000 as main, it's trash.
 
I would try:

- different firmware on AX11000 (firmware bug?)
- disabling AX on AX11000 (AX compatibility bug?)

At least to find where the problem is. Since you run stock Asuswrt you can report it to Asus.
 
I would try:

- different firmware on AX11000 (firmware bug?)
- disabling AX on AX11000 (AX compatibility bug?)

At least to find where the problem is. Since you run stock Asuswrt you can report it to Asus.
- firmware bug? Look online at asus.com - these are the latest firmware versions. I don't know why.
- I absolutely tried disabling AX - did not make a difference b/c backhaul has to use AC anyhow (low-5G i.e. channel 36, 40, 44, etc.)
(funny enough, even with repeater using the low-5G, I'm able to force my endpoints to use it, and get really good performance; 5G-high is still pretty crap on my phones and computers - AX or not)
My key question - dating back 2-3 years ago when I asked this re: RT-AC5300 as main (prior to scoring a GT-AC) was why it's so?
I've also tried literally 50+ endpoints - from iPhone 6 through 14 Pro, iPad 2 to iPad 11" Pro (M2), MacBooks from 2011 through 2022 models, Android 6 through 13, and Windows boxes (7, 10, 11) all over the spectrum.
And my desktops and work laptop are plugged into the repeater (now AiMesh node).
It's just... bizarre.
There appears to be an issue that is beyond my understanding. 400+ Mbps with a main unit released in 2015 (7+ years ago)... but in reverse (inverse?), only 90-110Mbps with a unit produced in 2018, with much more power and RAM.
 
- firmware bug? Look online at asus.com - these are the latest firmware versions

Wait to see 388 firmware... quite a few already reverted to 386 firmware on AX86U router. It has multiple minor GUI bugs, often crashing speed test and a major firewall bug recently discovered. What else is wrong is still unknown. Users are beta testers again with official Asuswrt releases.
 
Ideal placement 'stuff' does matter, yes, even inches make a difference. Each time you're setting up a different router/model.

What works for one model router doesn't necessarily mean it will work for another. This includes control channels too. It doesn't matter what channels are in use, it just depends on what channels the router sees...
 
Ideal placement 'stuff' does matter, yes, even inches make a difference. Each time you're setting up a different router/model.

What works for one model router doesn't necessarily mean it will work for another. This includes control channels too. It doesn't matter what channels are in use, it just depends on what channels the router sees...
Let me again explain - since i respect you, but we have difficulty communicating.
These are exactly the same two routers.
The only difference is which one is main ("master?") and node (slave"?).

When AX is master, performance is poor (the exception are computers wirelessly attached to the main).

The settings are, for all intents and purposes, standard. They are also the same in both orientations. I did not restore old configs. I do not use scripts. I do not mess with anything. Straight out of the box (I replaced my AX11000 since the first one got corrupted or fizzled out) - only change I make is my NAT subnet. That's it. Not even time zone (which I should) or Daylight Saving settings.

Distance does not matter. I moved the slave all over. I moved the master a couple of places. I had them so close that AiMesh faulted and attempted to use 2.4Ghz. Speed was within 2% in all places in which I moved the AX throughout my second floor.

One final note:
In production, I use an old AC68U as a "standard" repeater - and I have three of them I can warm-swap if it gets corrupted (happens 2-3x a year).
I find that the RSSI of the AC68U (any of the three) <> AC5300 (both GT and RT) is between -47 and -54.
However, the RSSI of AX11000 <> AC5300 is -65-70, and AC68U <> AX11000 is equally poor.

So... it could very well be that the AX11000 (or any WiFi 6 router?) is a horrible main when paired with AC routers - but all I appear to get is questions and arguments.

I posted this in case anyone else is beating themselves up.
 
Wait to see 388 firmware... quite a few already reverted to 386 firmware on AX86U router. It has multiple minor GUI bugs, often crashing speed test and a major firewall bug recently discovered. What else is wrong is still unknown. Users are beta testers again with official Asuswrt releases.
I got it - I'm using standard Asus firmware as if I'm a dumb-sht user who can't Google Merlin.
 
Asuswrt-Merlin won't help. Asuswrt bugs usually find their way in Asuswrt-Merlin releases. All AiMesh and Wi-Fi is closed source. Some AiMesh features are broken in Asuswrt-Merlin. With your wireless AiMesh - stay on stock Asuswrt. Otherwise your main router won't even see your node.
 
Asuswrt-Merlin won't help. Asuswrt bugs usually find their way in Asuswrt-Merlin releases. All AiMesh and Wi-Fi is closed source. Some AiMesh features are broken in Asuswrt-Merlin. With your wireless AiMesh - stay on stock Asuswrt. Otherwise your main router won't even see your node.
I’ve used and contributed to Merlin for years. But there’s no Merlin for the GT-AC so I wanted to match as best I could.
My AC-68U is on Merlin 386.7_2.
 

Similar threads

Latest threads

Sign Up For SNBForums Daily Digest

Get an update of what's new every day delivered to your mailbox. Sign up here!
Top