I have set up a Wi-Fi network based on two Belkin RT3200 and a Redmi AX3200, where two snapshots are installed with the same version (>=24.10).
Fast roaming works perfectly; however, with some devices (e.g., Pixel 9 Pro or PCs with Intel AX Wi-Fi), it seems as if there is a network cap that only disappears by restarting the wireless interface.
The behavior is the same across all three devices, and even disconnecting the device via Luci or directly from the device does not resolve the issue.
They still need "network" parameter to bridge wifi to a sw bridge and wire.
Probably some parameters still missing in the radio config as to set them to independent channels etc.
I just want to review them all through. As for old and stable devices - why not exactly 24.10 (snapshot or rc/release)
If it were a network problem the ip should not be there and if the wifi interface was restarted I should still not be surfing.
config interface 'loopback'
option device 'lo'
option proto 'static'
option ipaddr '127.0.0.1'
option netmask '255.0.0.0'
config globals 'globals'
option ula_prefix 'fd28:3e20:efe5::/48'
config device
option name 'br-eth'
option type 'bridge'
option igmp_snooping '1'
list ports 'lan1'
list ports 'lan2'
list ports 'lan3'
list ports 'lan4'
list ports 'wan'
list ports 'clients'
list ports 'iot'
config bridge-vlan
option device 'br-eth'
option vlan '200'
list ports 'wan:t'
config bridge-vlan
option device 'br-eth'
option vlan '201'
list ports 'iot:u*'
list ports 'wan:t'
config bridge-vlan
option device 'br-eth'
option vlan '254'
list ports 'clients:u*'
list ports 'lan1'
list ports 'wan:t'
config interface 'wan'
option device 'br-eth.200'
option proto 'dhcp'
The problem is also difficult to explain
I have a perfect signal (-55dbm) but it is as if I am being capped. In fact if I run a speedtest (iperf3 or speedtest app), the speed rises to about 20mbps and then drops to 0.1Kbps.
If I restart the wifi interface everything works (speedtest about 600Mbps) for a few hours.
Set pairs 36 52 , 36 100, 36 116 preferring dfs side
If any gets knocked out to 36 make it auto but limiting to remaining channels eg channes 36 51 116 channel auto.
That is the most commonly used 5GHz channel, so it is possible that there is congestion on that channel if there are lots of 5 GHz access points nearby.
After a few days when the wifi was working fine, the problem appeared again.
Even doing a few reboots the problem reappeared every 2/3 hours.
I carried out some tests by deleting the wireless network and adding it by only setting the enable for fast roaming, but nothing, the problem is still there.
config wifi-device 'radio1'
option type 'mac80211'
option path '1a143000.pcie/pci0000:00/0000:00:00.0/0000:01:00.0'
option band '5g'
option country 'IT'
option log_level '2'
option cell_density '1'
option he_su_beamformee '1'
option he_bss_color '64'
option htmode 'HE80'
option channels '36 100' #also without this setting problem appear
option channel '100' #other AP are on different channel/s
config wifi-iface 'wifinet1'
option device 'radio1'
option mode 'ap'
option ssid '<essid>'
option encryption 'psk2'
option ifname 'clients'
option key '<pwd>'
option ieee80211r '1'
option ft_over_ds '0'
option ft_psk_generate_local '1'
The only error I notice in the logread is this: daemon.err hostapd: nl80211: kernel reports: key addition failed
When problem appear to restore the wireless i need to restart the radio interface ubus call network.wireless down '{"device":"radio1"}' && ubus call network.wireless up '{"device":"radio1"}'
No way also with channel changes or with restore of an old configuration there is the problem.
how i can debug mt76 messages to better understand the situation?
No errors in your log, key of 11r is in kernel and needs no change. Check iwinfo after starting AP that it uses dfs channels, then check again if it ever runs slow.
I am currently having the EXACT same issue with a Linksys E7350 which is the Linksys version of the Belkin RT3200. It is configured as a dumb AP with firewall and dnsmasq disabled and 5 different SSIDs each bound to a different VLAN interface on the switch. One of the SSIDs performs at about 10% of the rest on download speeds from the internet (however strangely LAN to LAN speed tests perform as expected for 802.11ax). The SSIDs that are routed out to VPN endpoints suffer no performance penalty. I've been working on this issue unsucessfully. Dumb AP is running 23.05.5
I think there is a larger issue at play here. I've flashed this router with OpenWRT 24.10-rc5 and the issue persists and is getting worse since the WAN NIC now wants to autonegotiate to 100M at boot and won't establish a gigabit connection unless the wire is unplugged and replugged forcing renegotiation of the link. I've tested all the wiring with other equipment and this is not the issue; I was able to push the full gigabit with other eqipment.
I'm seeing a real issue with speeds on this unit now that I've not seen in the past and it has something to do with how VLANs are handled.
also with channel changes or with restore of an old configuration there is the problem.