Seek advice on how to troubleshoot wifi lost carrier issue

[bertlea]

I found my Pi-ZeroW got some wifi disconnection issue recently. It seems to have a 'carrier lost' event every 5 to 10 minutes. I do not have heavy traffic on the router nor the Pi. It is also very close to the router, just 2 feets away. It is running:

Code: Select all

Linux HN 4.14.34+ #1110 Mon Apr 16 14:51:42 BST 2018 armv6l GNU/Linux

The Pi was still running fine and seems still connected to the Internet via this router for days when this happening. Also, because there is a lower priority login (the guest account of the same router) configured in the wpa_supplicant.conf file, it finally switched to the guest wifi account by itself after about 3 days torture of this problem. This is when I realised an issue as the guest wifi account do not allow me to access from other systems to this Pi, but this Pi still able to send out message via the router.
Below is a small fragment of the log file (daemon.log.1) with ip and mac addresses masked off. Any advice what might went wrong or what I can check next?

Code: Select all

Jul 21 16:04:51 HN dhcpcd[251]: wlan0: no IPv6 Routers available
Jul 21 16:10:43 HN dhcpcd[251]: wlan0: carrier lost
Jul 21 16:10:44 HN dhcpcd[251]: wlan0: deleting address fe80::xxxx:xxxx:xxxx:xxxx
Jul 21 16:10:44 HN avahi-daemon[210]: Withdrawing address record for fe80::xxxx:xxxx:xxxx:xxxx on wlan0.
Jul 21 16:10:44 HN avahi-daemon[210]: Leaving mDNS multicast group on interface wlan0.IPv6 with address fe80::xxxx:xxxx:xxxx:xxxx.
Jul 21 16:10:44 HN avahi-daemon[210]: Interface wlan0.IPv6 no longer relevant for mDNS.
Jul 21 16:10:44 HN dhcpcd[251]: wlan0: deleting default route via 192.168.xx.1
Jul 21 16:10:44 HN dhcpcd[251]: wlan0: deleting route to 192.168.xx.0/24
Jul 21 16:10:44 HN avahi-daemon[210]: Withdrawing address record for 192.168.xx.yy on wlan0.
Jul 21 16:10:44 HN avahi-daemon[210]: Leaving mDNS multicast group on interface wlan0.IPv4 with address 192.168.xx.yy.
Jul 21 16:10:44 HN avahi-daemon[210]: Interface wlan0.IPv4 no longer relevant for mDNS.
Jul 21 16:10:44 HN dhcpcd[251]: wlan0: carrier acquired
Jul 21 16:10:44 HN dhcpcd[251]: wlan0: IAID eb:44:8f:48
Jul 21 16:10:44 HN dhcpcd[251]: wlan0: adding address fe80::xxxx:xxxx:xxxx:xxxx
Jul 21 16:10:45 HN dhcpcd[251]: wlan0: soliciting an IPv6 router
Jul 21 16:10:45 HN dhcpcd[251]: wlan0: rebinding lease of 192.168.xx.yy
Jul 21 16:10:45 HN dhcpcd[251]: wlan0: probing address 192.168.xx.yy/24
Jul 21 16:10:46 HN avahi-daemon[210]: Joining mDNS multicast group on interface wlan0.IPv6 with address fe80::xxxx:xxxx:xxxx:xxxx.
Jul 21 16:10:46 HN avahi-daemon[210]: New relevant interface wlan0.IPv6 for mDNS.
Jul 21 16:10:46 HN avahi-daemon[210]: Registering new address record for fe80::xxxx:xxxx:xxxx:xxxx on wlan0.*.
Jul 21 16:10:50 HN dhcpcd[251]: wlan0: leased 192.168.xx.yy for 43200 seconds
Jul 21 16:10:50 HN avahi-daemon[210]: Joining mDNS multicast group on interface wlan0.IPv4 with address 192.168.xx.yy.
Jul 21 16:10:50 HN avahi-daemon[210]: New relevant interface wlan0.IPv4 for mDNS.
Jul 21 16:10:50 HN avahi-daemon[210]: Registering new address record for 192.168.xx.yy on wlan0.IPv4.
Jul 21 16:10:50 HN dhcpcd[251]: wlan0: adding route to 192.168.xx.0/24
Jul 21 16:10:50 HN dhcpcd[251]: wlan0: adding default route via 192.168.xx.1
Jul 21 16:10:59 HN dhcpcd[251]: wlan0: no IPv6 Routers available
Jul 21 16:20:43 HN dhcpcd[251]: wlan0: carrier lost
Jul 21 16:20:44 HN dhcpcd[251]: wlan0: deleting address fe80::xxxx:xxxx:xxxx:xxxx

[wesleywitt]

+1

I'm seeing similar behavior on my RPI Zero W. The lost carrier happens several times per day, Most times it eventually reconnects buy can take 30 minutes to hours to eventually reconnect. Sometimes it will never reconnect and I have to power cycle the RPI. I've updated the firmware and disabled power mgmt on the WIFI chip.

Very unreliable and frustrating because I need to have consistent WIFI in my scenario. I'm considering trying a PI Zero (with no WIFI) and a USB dongle to achieve reliability. I used to have a RPI 2 with a dongle in this application and the WIFI connected was 100% reliable.

There are many, many posts everywhere about this issue but no real resolution that I can find.

[bertlea]

I spent sometime on it and I think I may found the cause of the carrier loss for my case. It seems my cheap router will go to sleep and causing the carrier loss of the connection every 5 to 10 min. if no network traffic goes to that router. Because I found the issue is always happening during the period that nobody is at home and all other internet connected device is powered off except my Pi-Zero W. Originally the Pi-Zero W will report it's own status every 10 min. Then the issue happens sometime. Now, after I changed it to send out it's own health status to the Internet every 1 min. No carrier loss observed so far. It is now running without issue for a week, so I think it is the time-out sleep issue of my router.

Hope it may help others if the issue happens in similar situation.

[btcruiser]

My plan is to use a Raspberry Pi Zero W as a process controller which means it has to be up and available for several weeks at a time. I was experiencing the same issues as listed in this topic, and I would have to echo the previous post that some access points turn their radios off during periods of inactivity causing the 'carrier lost' syslog entries. As indicated by bertlea just setting up some kind of traffic across the access point to keep the radio active will resolve the issue. In my case I have another wired Raspberry Pi perform pings every minute to the controller Raspberry Pi Zero W and there are no longer any syslog messages concerning wlan0 resetting. Most likely when the controller is fully working this will not be a problem as it will be using MQTT which has a 60 second keepalive that will force the access point radio to stay active.

[wesleywitt]

The pi w is too unreliable to be useful in any way. I've thrown all my pi w into the trash am using pi2s. They are 100% reliable all of the time.