WDS client sporadically drops connection (DEAUTH_LEAVING)

Hello everyone!

I am really desperate for help! I have a WDS repeater set-up between a WRT1200AC (client) and a WRT3200ACM (ap). The client is sporadically dropping connections every second day (night) or so. So far I have been restarting the client regularely, but I could not get it stable! I noticed that connection drops seemingly always happen at night, I don't know whether this could be a hint. Meanwhile I enabled logging, the WDS interface is wlan0. The logs for the drop are very specific:

Thu Aug  8 02:00:40 2019 kern.debug kernel: [189438.399284] ieee80211 phy1: Mac80211 start BA c8:85:XX:XX:XX:XX
Thu Aug  8 04:31:12 2019 kern.info kernel: [198479.900849] wlan0: deauthenticated from 60:38:XX:XX:XX:XX (Reason: 3=DEAUTH_LEAVING)
Thu Aug  8 04:31:12 2019 kern.info kernel: [198479.909312] br-lan: port 4(wlan0) entered disabled state
Thu Aug  8 04:31:12 2019 daemon.notice netifd: Network device 'wlan0' link is down
Thu Aug  8 04:31:12 2019 kern.debug kernel: [198480.236486] ieee80211 phy0: change: 0x40
Thu Aug  8 04:31:13 2019 kern.debug kernel: [198480.306388] ieee80211 phy0: change: 0x100
[...]

Notice that between 02:00:40 and 04:31:12 there was no activity on the router. After DEAUTH_LEAVING the client never recovers automatically! After I triggered a restart of the wlan0 interface on the LEDE UI, the client recovered instantaneously (I can attach the logs of that as well if needed). LEDE UI shows "Wirless not associated" for the interface when the connection dropped:

I am out of ideas why the interface doesn't recover itself automatically... any idea? Please also note channel 136 is a DFS channel in Germany. I am unable to switch to any lower (non-DFS) channels, because they are too crowded.

I also checked the AP, seemingly it restarted at arround that time! Thus the question remains: Why did the STA not reconnect automatically after the AP was back up? Is there anything I could do to force that?

Thank you & best regards,
Kristian

I think your problem same here.

Hello! Thanks for finding this!

Unfortunately I don‘t think it fits! It‘s not caused by a DEAUTH_LEAVING event, also I had this problem now for month on several different releases of OpenWrt! Right now both AP and STA are running on OpenWrt 18.06.4 r7808-ef686b7292.

I can now also reproduce the problem, by simply cutting the power of the AP, the client router never succeeds to reconnect, eventhough the AP beeing back up after 1-2 minutes (after waiting for the DFS scan). Any idea how to make the client auto reconnect to the AP, after it looses connection? If I just turn off the AP and restart it, the client logs simply say:

Thu Aug  8 21:05:39 2019 daemon.notice netifd: Network device 'wlan0' link is down
Thu Aug  8 21:05:39 2019 kern.info kernel: [258206.598893] br-lan: port 4(wlan0) entered disabled state

It never recovers automatically. Is it normal for WDS clients to act that way?

Thanks & best regards,
Kristian

i have compiled with last source and no problems

YouHua WR1200JS
Architecture MediaTek MT7621 ver:1 eco:3
Version du micrologiciel OpenWrt SNAPSHOT r10703-e1f588e446 / LuCI Master (git-19.217.64522-0cf9d53)
Version du noyau 4.14.136