Wifi bridge disconnects in regular intervals

I use two Netgear R7800 router with OpenWRT to bridge two network segments.
One router is in Master Access Point (WDS) mode, the second is in Client mode. This has worked since a couple of years. We had a longish (multiple days) internet interruption and since coming back up I have the wireless bridge getting interrupted in regular intevals, maybe once an hour, for a couple of seconds.

On each interruption I see the following in the dmesg of the client. Nothing is seen on the master

[ 1547.200671] wlan0: deauthenticated from bc:a5:11:31:f3:68 (Reason: 3=DEAUTH_LEAVING)
[ 1547.200882] br-lan: port 4(wlan0) entered disabled state
[ 1547.306077] ath10k_pci 0000:01:00.0: peer-unmap-event: unknown peer id 2
[ 1547.306127] ath10k_pci 0000:01:00.0: peer-unmap-event: unknown peer id 2
[ 1609.095389] wlan0: authenticate with bc:a5:11:31:f3:68
[ 1609.102692] wlan0: send auth to bc:a5:11:31:f3:68 (try 1/3)
[ 1609.105987] wlan0: authenticated
[ 1609.115006] wlan0: associate with bc:a5:11:31:f3:68 (try 1/3)
[ 1609.119415] wlan0: RX AssocResp from bc:a5:11:31:f3:68 (capab=0x11 status=0 aid=1)
[ 1609.122008] wlan0: associated
[ 1609.178106] br-lan: port 4(wlan0) entered blocking state
[ 1609.178153] br-lan: port 4(wlan0) entered forwarding state

How can I debug this ?

Have you tried the following:
Power down both routers
Power up the master
After the master is booted up wait three minutes and power up the client

Yes, I have powered both routers down, then started the master then a couple of minutes later the client.

I have also updated upgraded all installed packages (was not done since a while).

Is there a way I can get a more detailed log of what is happening ?

Could this be WPA rekeying?

How can I find out ?
Can I turn that off ?

I do not know if this applies to your situation but see:

I don't think this has to do with upgrading packages. The problem came up without any packages being upgraded and persisted after I did upgrade.

i was getting this frequently with my home hub 5a (no 23.x for this target yet & ath 10k radio), i was also seeing this on my TP-Link TL-WR2543ND, fairly often with 19.x & 21.x, didnt spent too much time testing 22.x but on 23.x RC3 the TP-Link ive stopped seeing deauth reason=3 messages, only issues i see now is beacon lost but that may be signal related (-70/-95 dBm) but TP-Link seems to have improved wifi stability on 23.x for me (and it has a 3x3 2.4Ghz radio)

So you suggest I upgrade from 19.x to 23.x ?

worth a go, I've upgraded and downgraded several times for testing, trying to maybe find a more stable Wi-Fi version (just using sysupgrade images) do just backup your confing and if 23 is a worse experience should be able to return to 19 and get back up and running fairly fast, i also turn on "Disable Inactivity Polling" and turn off "Disassociate On Low Acknowledgement", this didn't make much difference in older versions if i remember and may not even apply to client mode but just something i do automatically now from previous experimentation :slight_smile:

EDIT: you will probably want to clear your settings during upgrade & downgrade, alot of devices are moving to DSA and these config files are different i think

Yep, just upgraded the client to 22.03 (latest stable). Was quick and painless :-).

Will see how it runs. Needs a couple of hours to be sure.

1 Like

The problem is still here; just the log looks different:

[ 1638.667356] wlan0-1: authenticate with 1c:ed:6f:70:47:41
[ 1652.124015] ath10k_pci 0001:01:00.0: mac flush vdev 1 drop 0 queues 0x2 ar->paused: 0x0  arvif->paused: 0x0
[ 1667.204066] ath10k_pci 0001:01:00.0: mac flush vdev 1 drop 0 queues 0x2 ar->paused: 0x0  arvif->paused: 0x0
[ 1682.284077] ath10k_pci 0001:01:00.0: mac flush vdev 1 drop 0 queues 0x2 ar->paused: 0x0  arvif->paused: 0x0
[ 1697.364108] ath10k_pci 0001:01:00.0: mac flush vdev 1 drop 0 queues 0x2 ar->paused: 0x0  arvif->paused: 0x0
[ 1712.454331] ath10k_pci 0001:01:00.0: mac flush vdev 1 drop 0 queues 0x2 ar->paused: 0x0  arvif->paused: 0x0
[ 1742.624212] ath10k_pci 0001:01:00.0: mac flush vdev 1 drop 0 queues 0x2 ar->paused: 0x0  arvif->paused: 0x0
[ 1757.694224] ath10k_pci 0001:01:00.0: mac flush vdev 1 drop 0 queues 0x2 ar->paused: 0x0  arvif->paused: 0x0
[ 1764.837774] wlan0-1: authenticate with 1c:ed:6f:70:47:41

Oh, i havent seen that one that i remember before, seems like a more useful output into the log (although beyond me) must try 22 on my hub5 again and see, seems like its ath 10k issue :slight_smile: