wlan0 issue in applience-mode

[rehflux]

Hello and thanks for the great work, comitup is just what i was looking for.
Everything is running as desired, but in appliance mode wlan0/ap is restarted every 6 minutes, which leads to connection interruptions for the clients.

Here is an excerpt from the logs:
Jul 22 15:15:58 maximegalon wpa_supplicant[430]: wlan0: interface state ENABLED->DISABLED
Jul 22 15:15:58 maximegalon wpa_supplicant[430]: wlan0: AP-DISABLED
Jul 22 15:15:58 maximegalon kernel: ieee80211 phy0: brcmf_vif_set_mgmt_ie: vndr ie set error : -52
Jul 22 15:15:58 maximegalon kernel: ieee80211 phy0: brcmf_vif_set_mgmt_ie: vndr ie set error : -52
Jul 22 15:15:58 maximegalon wpa_supplicant[430]: wlan0: CTRL-EVENT-DISCONNECTED bssid=e4:5f:01:f7:33:bd reason=3 locally_generated=1
Jul 22 15:15:58 maximegalon comitup[4310]: OK
Jul 22 15:15:58 maximegalon NetworkManager[372]: <info> [1690031758.5411] device (wlan0): supplicant interface state: completed -> disconnected
Jul 22 15:15:58 maximegalon NetworkManager[372]: <info> [1690031758.5413] device (p2p-dev-wlan0): supplicant management interface state: completed -> disconnected
Jul 22 15:15:58 maximegalon comitup[4312]: OK
Jul 22 15:15:58 maximegalon wpa_supplicant[430]: Using interface wlan0 with hwaddr e4:5f:01:f7:33:bd and ssid "BabyWifiVpn"
Jul 22 15:15:58 maximegalon wpa_supplicant[430]: wlan0: interface state UNINITIALIZED->ENABLED
Jul 22 15:15:58 maximegalon wpa_supplicant[430]: wlan0: AP-ENABLED
Jul 22 15:15:58 maximegalon wpa_supplicant[430]: wlan0: CTRL-EVENT-CONNECTED - Connection to e4:5f:01:f7:33:bd completed [id=0 id_str=]
Jul 22 15:15:58 maximegalon NetworkManager[372]: <info> [1690031758.6442] device (wlan0): supplicant interface state: disconnected -> completed
Jul 22 15:15:58 maximegalon NetworkManager[372]: <info> [1690031758.6443] device (p2p-dev-wlan0): supplicant management interface state: disconnected -> completed

Is this behavior intended? If not, is there a known fix?
Thank you!

[rehflux]

Update and additional infos about my problem:
Running on RPi4B with additional USB-Wifi (RTL8812BU-Chipset, driver: https://github.com/morrownr/88x2bu-20210702)

For me it seems like the "problem" is caused in states.py in this section:

@timeout
def connected_timeout(dummy: int) -> None:
    active_ssid: Optional[str]
    active_ssid = nm.get_active_ssid(modemgr.get_state_device("CONNECTED"))
    log.debug("connected_timeout comparing {} to {}".format(connection, active_ssid))
    if connection != active_ssid:
        log.warning("Connection lost on timeout")
        dev = modemgr.get_state_device("CONNECTED")
        set_state("CONNECTING", candidate_connections(dev))

    if modemgr.get_mode() == modemgr.MULTI_MODE:
        wpa.check_wpa(modemgr.get_ap_device().Interface)

        active_ssid = nm.get_active_ssid(modemgr.get_state_device("HOTSPOT"))
        if not active_ssid:
            log.warning("Hotspot lost on timeout")
            set_state("HOTSPOT")

In verbose mode comitup.log says "connected_timeout comparing SSID to SSID", where both SSID is my home-wifi-ssid. Then it seems the wpa.check_wpa restarts the AP. If i comment the line wpa.check_wpa(modemgr.get_ap_device().Interface) out, everything runs well.

[davesteele]

Sorry for the delay.

This is a hack I added 4 years ago. AP mode was eventually failing, I could not reliably detect the failure, and restarting wpa_supplicant fixed the problem. The call you disabled causes wpa_supplicant to be restarted every 5 minutes.

It's better if the application didn't have to do that. I'd be interested to know if you have any issues with the AP staying up more than a couple of days.

Note - I've noticed that I am getting kicked off of my Zero running in appliance mode, but it is less than once every 5 minutes. So it is not every time.

[rehflux]

It's better if the application didn't have to do that. I'd be interested to know if you have any issues with the AP staying up more than a couple of days.

I didn’t restart the Pi for 3 days, I was able to connect to the AP, but the connection was very slow and seemed to have high latency. After a restart everything was okay again
I have now swapped the WLAN devices (WLAN1 = AP now; the WiFi USB stick can AP), let’s see how the problem behaves now.

Note - I've noticed that I am getting kicked off of my Zero running in appliance mode, but it is less than once every 5 minutes. So it is not every time.

I have also noticed this behavior, but for example, SSH connections were interrupted every time.

[davesteele]

Note to future self - the problem addressed by f5b2686 remains, at least after a couple of days.