My second GL-AX1800 arrived today. I see that the 20% pricing glitch has been corrected meanwhile.
I already bricked and debricked the device 3 times. https://docs.gl-inet.com/router/en/3/tutorials/debrick/ was helpful. On the first time, I simply ran sysupgrade from the command line via SSH. On the second and third time (upgrading from openwrt-ax1800-4.6.8-1017-1729176012.img), I used the web interface, and got the same thing. I was unable to enable any failsafe mode either.
The symptoms that I am seeing is that the LED stays solid blue and the router can’t be accessed via any wired Ethernet port. There is no DHCP server responding on the LAN ports nor any DHCP requests being sent to the WAN port. I suspect that something was broken between 9f76070e2de5c28830ede70c31fb81278f1c7183 (May 27) and today’s snapshot. I’ll try to install an older sysupgrade firmware snapshot.
Edit: I figured out my mistake. My self-built firmware worked, but so did the factory.bin from the firmware-selector as well. In the manufacturer’s web interface for an upgrade from a local image, both factory.bin images were identified as "SNAPSHOT", while the "sysupgrade" image was identified as "unknown".
Yes, I got it working to the extent I needed for now. I followed https://openwrt.org/docs/guide-user/network/wifi/wifiextenders/bridgedap, disabling the DHCP client and server. I made the WAN port part of the LAN bridge, although I could as well have connected the main router via one of the LAN ports. I configured it via SSH, because for some reason, LuCI does not run at all with yesterday’s snapshot firmware from the firmware-selector.
Curiously, the WiFi of a 2016 laptop achieved about 20 Mb/s according to speedtest.net. With a 5-year-old flagship mobile phone I initially got only 10 Mb/s; on some runs it exceeded 100 Mb/s. The devices were located upstairs 1 to 2 meters from the second router. I used identical WLAN configuration for both routers, which I realize could have caused severe interference. Today I repeated the experiment with only the router in the basement running, and the devices located upstairs. The throughput was similarly bad (but still OK for our general use). Via wired Ethernet through the 2 routers, the performance was as expected, throttled by my ISP to a bit over 200 Mb/s.
I did not experiment with https://openwrt.org/docs/guide-user/network/wifi/roaming yet. I think that for starters, I must use non-overlapping wifi channels. I’d also like to experiment with VLANs and VLAN specific WiFi SSIDs at some point.
could someone who owns an AX/T/1800 run the Wireguard performance test on it, and post the results in A Wireguard comparison DB ? test setup is described in the 1st post.
I experimented with IEEE 802.11r WiFi roaming this weekend. I was impressed how smoothly it worked. I configured non-overlapping radio channels for both 2.4 GHz and 5 GHz WLAN bands and enabled 802.11r and FT over DS.
I got the best results with a setup where there was a few meters of overlap between the main router in the basement and the second router in a garden shed, which were connected to each other by gigabit Ethernet. The handover would take place near the house door. Initially I had set up the second access point indoors upstairs, but I had trouble forcing any handover, because each router would easily cover the entire house on their own.
With the two laptops that I tested (Lenovo ThinkPad X220 from the early 2010s, and a 2016 laptop with Skylake and Intel 8260), the handover was very smooth, with less than 100ms of delay. I tested this by opening an SSH session from the laptop to one of the routers and executing ping -A 192.168.1.x (the address of the laptop) on it. The SSH session survived during the handover, and the round-trip time reported by ping typically was less than 2 ms, and even during the handover it was less than 100 ms.
On an Android phone, the connection would be interrupted for up to 10 seconds during the handover. A WiFi based Signal voice call was disturbed but not disconnected during the handover. It is possible that I had configured a mismatching mobility domain between the two routers; I did not double check this before tearing down the experiment.
One low-end Android phone from 2016 would fail to authenticate with WiFi as long as 802.11r on the 2.4 GHz interface was enabled.
I've been using WiFi roaming between the GL-iNet AX1800 and a couple of B1300s for a couple of years. The advice I had heard was to use adjacent radio channels as that helps devices to notice a stronger alternative. Modern Android will actively switch to a better network. I also enable usteer and had previously used Dawn. Configuration for usteer is not well documented and I'm not sure how well it works in practice. It's also a good idea to lower the transmit strength of access points as that makes it less likely that they get used from a high distance. I find handover to be much quicker than 10 seconds. I barely notice interruption at all. There are a few other things to enable like 802.11k and replace wpad-basic-mbedtls with wpad-mbedtls.
I've also successfully used VLANs on the AX1800 for a long time. The configuration for that changed when moving to the newer release - after configuring Bridge VLAN Filtering, it is important to change items such as "lan" under interfaces to use br-lan.1 or whatever the appropriate VLAN is instead of just br-lan.
I updated to a fresh snapshot as recently as last Thursday and that's been working fine.