question

I used SSH to log into the Cerbo and discovered evidence that wifi freezing occurs often. So the shore power brown out was probably unrelated and just observation bias on my part.

When the cerbo wifi freezes, I can go into my Peplink router and manually kick the Cerbo off its wifi connection. This triggers Cerbo's wifi reconnection logic (without rebooting) upon which VRM is updated again. Otherwise, wait for the cerbo to hit its VRM 'publishing failed' timeout and reboot itself. The Cerbo's RSSI is -38db (very strong signal), which make sense since it is only about a half meter from the access point. So this is not a case of a weak signal. I've also experimented with disabling suggested wifi channels (hdmi interference) which does not appear to have an effect although difficult to directly measure.

I will try a supported USB WiFi dongle to see if that is more reliable than the built-in chip & software.

do you have the cerbo with a SD card plugged in? I ask as when the Cerbo GX losts contact with the VRM it saves the data in the SD and when the connection is restored it catches up so you wont loose anything in the VRM portal. It doesn't mind if it lost connection to the WIFI or not.

a) if the WIFI keeps dropping I'd rather use the RJ45 cable to the router if it is so close then you can troubleshoot with confidence it's not a connectivity issue but something related with the when going from shore to inverter power.

b) I re-read your last message and if you are able to ssh into the Cerbo I'd automate a script that runs every 30 minutes or so you can turn off the wifi off and then on again.

Edit the crontab then add the line below, this will restart the network every 30 minutes.

crontab -e

*/30 * * * * /etc/init.d/network restart

I'd still test with a) first then going with b.

Thanks for your input @ffvictron. My ethernet port is consumed by a connection to a marine MFD (which insist on hosting its own dedicated network segment). So I cannot connect via hardwire as would be preferable. I have an SD card for the Cerbo next time I visit. Meanwhile, I am using VRM alerts to manage a response within 24-48 hours so as not to lose data.

I'll experiment with a periodic restart of the wifi as you suggest, at least until I have added an external Wi-Fi dongle. Good idea.

A mature, stable system should not require periodic restarts. Right now the log files do not give any failure indication before the Cerbo becomes unresponsive. To continue troubleshooting, I have been investigating how I might turn up debug logging levels for connmand (still have to figure out how to do it). Perhaps it is throwing a relevant error or at least providing a clue as to its state at the moment it becomes unresponsive.

@John Leslie: I dont have a Venus OS system to test but the logs should be in the same standard location as for the Unix systems or around : /var/log. I'd suggest checking the following for clues:

/var/log/messages
/var/log/wpa_supplicant.log
/var/log/dmesg