without seeing your config nobody will reply to this rant

…my near 20 years of using Mikrotik products…

I literally want to take this hAP ax³ out side and shoot it with my 12 gauge

what frequencies are best?

would you kindly pass along your setup?

See my apology and attached script in the original post. Login details are posted there as well.

I made them up on the fly just encase someone could not get into the settings I provided. I would never post any real credentials I use. Same with the password on the wifi security.

I will give it a try and report back. I find that curious about the bridge because I am sure I only used the "All" setting. I am to lazy to add them one at a time.
DHCP relay = is not needed since the server is on the same sub net.

If you are using it as an AP, you can bridge the WiFi interface to the LAN interface, so the AP will act as a switch.

Maybe you should specify vlan 1 in data path in the WiFi config, or some other vlan depending on what you use. Mikrotik uses vlan 1 as default.

If you are using it as an AP, you can bridge the WiFi interface to the LAN interface, so the AP will act as a switch.

That is entirely correct. However, that is also already the case, see config =P

Maybe you should specify vlan 1 in data path in the WiFi config, or some other vlan depending on what you use. Mikrotik uses vlan 1 as default.

Setting vlan to 1 explicitly is not required, and IMO it is better to leave VLAN-related settings alone in a VLAN-less environment, which this one seems to be.

How do you know it's VLAN-less ? Maybe the router is using VLAN 1. Most mikrotik gear will use VLAN 1 if left unconfigured, but will not mark packets as VLAN 1 for WiFi.

...my iPhone 14 Pro can connect to the 2.4Ghz side and surf the internet no issues...

How do you know it's VLAN-less ? Maybe the router is using VLAN 1. Most mikrotik gear will use VLAN 1 if left unconfigured, but will not mark packets as VLAN 1 for WiFi.

Because I read the config and there's no mention of VLANs?

Sure, the interface on the CRS that's connected to the ax3 could somehow configured as a VLAN1 access port. With somehow only tagged packets being admitted and untagged packets being dropped as opposed to tagged (Jackie Chan meme).

But that would mean that the entire ax3 would essentially be offline, and not a single client connected to it would be able to access the Internet. Which is not the case:

...my iPhone 14 Pro can connect to the 2.4Ghz side and surf the internet no issues...

/interface wifi set [ find default-name=wifi1 ] channel.band=5ghz-ax .frequency=5180,5240,5745 .width=20/40/80mhz configuration.country="United States" .mode=ap .ssid=House5GHz dis
abled=no name=House5GHz security=WifiSec

It works normally on the 2.4Ghz radio, pages loading very fast all good. Switching over the 5Ghz side, produces the almost the same result. Pages do load however they on average always take over a full minute to do so. So an improvement of sorts...

/interface/wifi/radio/reg-info country="United States" number=0

/interface/wifi/monitor [find default-name=wifi1]

• Try testing the speed really really close to the router, no more than 3 feet away.
• Try replacing the antennas, if you have different ones. Take care to switch the router off before unscrewing anything.

Here you go Tangent:

• …the default configuration, but instead this heavily-changed AP bridge thing you're trying to set up. While it is highly useful to see what you're trying to do, I want you to realize that the reason I asked for the defconf was so that I could have a clean baseline to diff my local configuration against, since it's closer to stock than yours is. Without that baseline, I was left with two hard choices: either try to produce a 3-way diff without the common baseline version on hand, or take the risk of resetting my own working router for the purpose of this comparison.
• …terse mode, which meant I had to go in and fix all the whitespace changes to get a clean diff
• …cleaned of PII; this is important not merely for your privacy but because the baselines I'm using here to diff against are cleaned of PII, causing more work in identifying the important differences. (For one, you've given away your router's serial number, which means a malefactor knows where to try breaking in if you were to ever enable MT's nebulously-named (hah!) Cloud service.)

• Naming the SSIDs differently for 2.4 and 5GHz isn't a great practice. It prevents roaming, which modern WiFi devices expect to handle on their own.
• Abstracting common elements like "wifi security…name=WifiSec" is a good practice. I'd recommend doing more of that with "/internet wifi configuration" and such. Ideally, the 2.4 and 5GHz sections should be identical but for radio-specific details, as in my (working!) config below.
• I find it worth repeating that setting channels on a home router is silliness unless you're out in the middle of nowhere or somehow able to control what your neighbors are doing. The only exception I find to this is to exclude one of the defaults where it's known to interfere, and even then, you can expect the radio firmware to figure that out on its own.
• The "interface=*B" bit is a result of configuring something one way, deleting the referent, and leaving the old config change behind, pointing at nothing. Remove it.

/interface wifi configuration add channel.skip-dfs-channels=10min-cac country="United States" disabled=no mode=ap name=HomeWiFi security.authentication-types=wpa2-psk,wpa3-psk .ft=yes .ft-over-ds=yes .passphrase="**ELIDED**" ssid="Home WiFi" 
/interface wifi set [ find default-name=wifi1 ] channel.band=5ghz-ax .width=20/40/80mhz configuration=HomeWiFi
/interface wifi set [ find default-name=wifi2 ] channel.band=2ghz-ax .width=20/40mhz configuration=HomeWiFi

...my iPhone roams between the 2.4 and 5GHz networks without hassle. I'm not the type to dive deeply into something that's working, so perhaps there is some strange handoff problem I don't see, but the important bit is that I don't see it. From my perspective, everything works fine, which is all I asked of it.

• If you just look at the config itself, rather than gawping at the red in the diff, you'll see that the original config provided by @webequipped is pretty much just default CAP config, with a few things changed around, with a lot of changes being not relevant to the situation at hand. You're not telling me setting NTP servers is gonna impact your wifi performance, now, are you?
• Having 2.4 GHz and 5 GHz networks separate is a matter of preference. Your clients aren't necessarily gonna be the smartest on earth. I find the roaming decisions of mine pretty dumb some of the time, despite all my networks being chef's kiss level optimized for roaming. Naming 2.4 and 5 GHz networks differently provides you with the ability to fine-tune your client's behavior, which I quite like.
• Having channels specified makes a heck of a lot of sense in a home environment. MikroTik routers are dumb with channel auto-selection, and if you think otherwise, you've just been moderately lucky. They will randomly select 5745/Ceee with max power set to 14 dBm by local regulations in Europe. They prefer higher 5 GHz channels, so if you have any devices that don't work with 5735-5815 frequency range, no 5 GHz for you for random periods of time. They will gaslight themselves into believing that 10-min DFS channels are really the bestest choice there is, and patiently wait for a radar event while you're trying to log into them to figure out why there's no wifi after you rebooted the router. (that, at the very least, can be avoided with the 'skip-dfs-channels' setting) And if your interference situation is good, which it might as very well be, setting channels explicitly in a small CAPsMAN setup makes a lot of sense. You just have one router on the non-DFS supported-by-every-device 5180 frequency, and another one on some wild shiny new 5745 that allows more power in some countries.
• All of this is irrelevant. None of this matters. The 2.4/5 GHz names aren't the problem. The channel selection isn't the problem. And the in-depth diff analysis you did doesn't actually matter. Because if you just scrolled down, you would see the updated config. With most of the config bloat removed. Without the "*B" dead pointer. Without the "interface=all" in bridge. And guess what? It kinda works now, but the problem is different. The problem now, is the old one. The one the ax3 seemingly can't escape...

I cleaned up the bridge, rebuild it from scratch, manually added each interface. I also removed the DHCP relay as the DHCP server is on the same network so it is not needed.

I now have it back in play and decide to test it again with my iPhone. It works normally on the 2.4Ghz radio, pages loading very fast all good. Switching over the 5Ghz side, produces the almost the same result. Pages do load however they on average always take over a full minute to do so. So an improvement of sorts...

I did order a new hAP ax3 and it arrived late last night. I now have that hAP ax3 manually configured with the same settings and replaced the other hAP ax3. Unfortunately no change in the behavior.

Here is the export from the new hAP ax3:

Code: Select all

# 2024-07-20 09:20:11 by RouterOS 7.15.2
# software id = removed 
#
# model = C53UiG+5HPaxD2HPaxD
# serial number = removed 
/interface bridge add name=LanBridge
/interface ethernet set [ find default-name=ether1 ] name=LAN1
/interface ethernet set [ find default-name=ether2 ] name=LAN2
/interface ethernet set [ find default-name=ether3 ] name=LAN3
/interface ethernet set [ find default-name=ether5 ] name=LAN5
/interface ethernet set [ find default-name=ether4 ] name=LANr4
/interface wifi security add authentication-types=wpa2-psk,wpa3-psk disabled=no encryption=ccmp,gcmp,ccmp-256,gcmp-256 group-key-update=1h name=WifiSec passphrase=removed wps=dis
able
/interface wifi set [ find default-name=wifi2 ] channel.band=2ghz-ax .frequency=2412,2432,2462 .width=20/40mhz-Ce configuration.country="United States" .mode=ap .ssid=House disable
d=no name=House2.4GHz security=WifiSec
/interface wifi set [ find default-name=wifi1 ] channel.band=5ghz-ax .frequency=5180,5240,5745 .width=20/40/80mhz configuration.country="United States" .mode=ap .ssid=House5GHz dis
abled=no name=House5GHz security=WifiSec
/interface bridge port add bridge=LanBridge interface=LAN1
/interface bridge port add bridge=LanBridge interface=LAN2
/interface bridge port add bridge=LanBridge interface=LAN3
/interface bridge port add bridge=LanBridge interface=LANr4
/interface bridge port add bridge=LanBridge interface=LAN5
/interface bridge port add bridge=LanBridge interface=House2.4GHz
/interface bridge port add bridge=LanBridge interface=House5GHz
/ip neighbor discovery-settings set discover-interface-list=!dynamic
/ip dhcp-client add interface=LanBridge
/ip dns set allow-remote-requests=yes
/system clock set time-zone-name=America/Los_Angeles
/system identity set name=AP2-AX-House
/system note set show-at-login=no
/system routerboard settings set auto-upgrade=yes

I am bewildered and dismayed to say the least....

Having 2.4 GHz and 5 GHz networks separate is a matter of preference.

MikroTik routers are dumb with channel auto-selection…max power set to 14 dBm by local regulations in Europe.

if you just scrolled down, you would see the updated config.

You're telling me that bridging the two wifi networks has no effect on the original problem statement, where one side works and the other doesn't?

I repeat: my config works. I just got a 656 Mbit/sec iperf3 result across my ax³ to an iPhone while standing right next to it. I get about half of that in the next room over. I've seen better, but while this may be slower than I'd like, it isn't "slow" by any reasonable standard.

Yes, it's true, I did not read each and every message in the thread after the one directed at me. I am clearly a very bad person and should stop trying to help.

naming SSIDs…might hide the problem, by the means of clients learning to roam away from the slow 5GHz radio

a hardware problem

you wrote a whole wall of text about an old config, probably spending, like, 10-30 minutes on it

1. I have the international version of the hAP ax3.
2. It auto-chose frequency 5745.

I had to dangle an RB1100 from an Ethernet cable

quick and dirty speedtesting action

But it is, in all likelihood, not with the ax3's configuration.

I have the international version of the hAP ax3.

It auto-chose frequency 5745

putting only one PoE port on it, and making it the same one for in and out.

<...>

I concede the usefulness of PoE out on this box for powering an LTE modem or WISP CPE antenna, but PoE in should be on ether5. I've asked, and nobody has given me an example of WAN CPE that provides PoE. That's a LAN-side function.

If you were trying to make us jealous by showing off your symmetric Internet connection, you've succeeded. I've gotta use iperf3 inside the house to stress this router's capabilities.

How do you make it admit that truth, please? All I can get is a long list of available channels, plus the "Scan" function, which doesn't show "me" in the list.

Mine's the "-US" variant even though it came to me through Getic.

'monitor' command from cli.

So you'll be able to test if the config works on the US version then

I also have a pet theory that some of these weird ax3s are coming from Amazon returns rebadged as new routers.

There is a chance that your cooked-radio hypothesis explains why it isn't as fast as the <NAUGHTY-U-WORD-EXPURGATED> router it replaced

I've taken enough risks with my home IT core for one Saturday. Maybe tomorrow, but probably not, and definitely not during the week; I work from home.

/interface wifi set [ find default-name=wifi1 ] channel.band=5ghz-ax .frequency=5180,5240,5745 .width=20/40/80mhz configuration.country="United States" .mode=ap .ssid=House5GHz dis
abled=no name=House5GHz security=WifiSec

Please do a simple test, use only 20 MHz width on 5 GHz, set frequency as 5180-5650. Set tx power 15. Disable 2.4 GHz. Be about 10 feet from the router.

The reason I suggested this is because my own ax3 shows worse performance on 80 MHz than on 40 MHz. Even 20 MHz performs better than 80 MHz, but 40 gives the best results to me. Maybe we received bad hardware.

>Remove the DHCP client

Is it okay to have this on the bridge - mainly so you can see what IP the access point/switch is using?

I must admit, that does make sense. In practice, the amount of sense it makes is vanishingly small, since AP can't route packets anywhere but back to the bridge anyway. But this is in the spirit of theoretical advice of removing variables that could affect the results of experiments, yes.