Some wifi clients lose internet connectivity while others don't

Hi, I'm taking care of the network at my sister's home and have a weird problem with it that I don't know how to debug properly. The network consists of 2 wifi access points in a single-family house. The ceiling on the ground floor is made out of ferroconcrete or something and makes it really hard for a single access point to broadcast the signal between the ground floor where the living room is and the second floor where the sleeping room and the children's rooms are. So there are 2 wifi access points installed in that house, one per floor and they are directly connected via wired ethernet.

One access point is a gl-inet GL-AR150 acting as an access point and the other one is a TL-WDR4300 acting as the firewall/router/bridge to the internet in the network. Both are running the latest version of OpenWRT that is available for the respective platform.

The symptom of the problem is that the samsung smartphone of my sister and the iphone of her son lose internet connectivity frequently. It seems they drop out of the network somehow in a way that causes the client device to report that it is still connected to the wireless network but lose the ability to communicate on layer 3. They respond to ICMP echo request while connected and become unresponsive to echo requests when they appear to have fallen out of the network.

Not all devices in the network are affected by this. Her daughter's iPad, my Oneplus Phone and my Notebook are fine though and they stay connected and responsive to ICMP echo requests. It also seems like it only happens when they are on the ground floor and not when they are on the second floor which means it appears to be a problem with the GL-AR150 on the ground floor. So I replaced that device with a spare TL-WA901NDv2 that is also running the latest OpenWRT version but the symptoms persists. There is nothing written to the system logs accessible via logread at all when this happens.

So I'm at a loss here for now and don't know how to further debug this problem. Any kind of help would be greatly appreciated.

general approach...

  1. lower the power on the upper level device
  2. scale back the wifi defaults ( Mhz bands / freq-supp-mode ) search forum
  3. see if no ipv6 or static ipv6 setting on the device alters behavior
  4. change the advanced android settings re: link detection
  5. roll back to previous release / forward to snapshot ( bottom floor device )

i'd be surprised if that doesnt yeild results... 1 might disrupt other devices so try with your spare ap or during scheduled downtime... consider trying 4 before 1... less end user interruption...

based on your symptoms... perhaps the driver for the ground floor chipset is flakey / incompat with that device... which may relate to key negotiation intervals etc. might be worth a search on that chipset and similar time based connection troubles... in a nutshell... a more aggressive/alternate policy on both ap may reveal symtomatology / results. or option 4 could pinpoint lower level driver bugs related to this...

report back if no progress there are more detailed options. if you do... let us know which band these devices seem to connect to the most...

2 Likes

Does this issue only occur with Android devices?