CerboGX: VRM Connection Error #153 on Starlink WiFi

Lately my CerboGX is failing to connect to VRM every time I turn on our Internet. The Wifi works fine and Cerbo is able to use it but only after manually disconnecting and reconnecting the Wifi network in Cerbo menus. I have to do this every day (our Wifi is a Starlink router, run off the inverter).

It seems the Cerbo gets stuck in a DNS name resolution problem. This error has been posted by others previously in the old forum. Is this a Victron bug? Toggling HTTPS off doesn’t help, then I just get the error from the HTTP library instead of HTTPS.

All other Internet functions are working over this wifi network, on multiple devices, with no special DNS settings at all.

The Cerbo is on firmware 3.52. I’ve tried a reboot already.

From the cerbo manual

Error #153
Connection problem, and then specifically an SSL related issue, such as in below screenshot: check the date and time setting of the Gx Device, and also the time zone. And check that your router is not showing a special disclaimer, login or acceptance page, like often seen in airports, hotels and other public wifi.

153 covers a host of issues. It is usually related to routing which is a network problem.

(So no or slow internet, intermittent internet router not passing it on for some reason.)

Hi, the date/time and timezone are correct, and the router doesn’t have a login page or interstitial of any sort.
Yes, it is a network problem with the Cerbo. The fact that disconnecting and reconnecting the wifi makes it work demonstrates that the wifi works fine.

Starlink wifi doesn’t have working Internet immediately upon boot, so it’s normal that name resolution would fail in the initial few minutes of the wifi network appearing. But that is true of many wifi situations, where Internet connectivity might not always be available.

It seems like what’s happening is the Cerbo gets that initial network request failure, and then simply never tries again. Or it has cached the initial name resolution error and never reattempts that part. Is there a way to fix this?