Further testing:
a) had my Chuck Norris probe running only every 30mins
b) Homey Pro 2023 web-app went offline, Homey Pro 2019 stayed accessible
c) My local NAS still received syslog event results from chucknorris API every 30mins. So Homey Pro 2023 obviously was still happy with respect to local-local connection as well as Homey-to-Internet.
d) By chance I saw that my NAS had reports in its event log that it had many nightly 3-6h updates of my dyndns settings. Strange, because usually this should only be needed upon IP updates. Stranger, the IPv4 logged was the same despite having a few dyndns updates in a row within second. Hard to tell, if this was due to IPv6 updates which I cannot see in the log, if that was the reason.
e) I tried something… I deactivated IPv6 support completely in my AVM fritzbox mesh master. It had to reconnect IPv4, and did not connect IPv6 as being disabled.
f) Homey Pro 2023 webapp was accessible again - No reboot of Homey Pro 2023 needed. It just was brought back to live… Either that was due to the general DSL reconnect or removing IPv6 out of the equation did the trick. At least sometimes I got the impression that a router restart instead of a Homey restart also could fix the connection.
Either that’s pure coincidence or is has something to do with IPv6. So next round of testing… Leave IPv6 deactivated in my router, leave everything else unchanged, and… wait…
Wouldn’t be the first device having trouble with IPv6.
I have a Fire TV stick that had internet disconnects (2 other Fire TV sticks never had this issue, but there is one - even being LAN connected, that freaks out and I never understood why this device of more than thirty others sporadically goes havoc in a quite similar fashion my Homey Pro 2023 does…). It was fine for weeks, but - and that could again by pure coincidence - yesterday also required some love, i.e. unplugging/rebooting.
So probably also something to try as a workaround… If you have IPv6 one could try if having only IPv4 helps, too. No guarantees, of course.
Updated 25.1.:
Just checked for CGNAT. I don’t have it. My router’s WAN IP is equal to what whatismyip.com says for my public IP. Also a tracert to my public IP does show just 2 entries: local IP of my DSL router and public IP.
However I can see that ~5mins after nightly DSL reconnects happening between 3-4 o`clock and usually around 3:2x I find ECONNREFUSED, EHOSTUNREACH, EHOSTUNREACH, ECONNREFUSED, Bad Gateway, Bad Gateway while doing a local request from Homey to another device (solar batterie’s webapi) sitting in the same network. These requests happend every 15sec and after 6-7 calls ~5min after the router’s DSL reconnect it get’s back to normal… However trying to use the web-based Homey app after that leaves my with “Homey is offline”. That is even if my WAN IP stays the same after the nightly reconnect which can but does not always happen.
Let’s see if Athom support having those logs and info can shed more light into that. Somehow it correlates with DSL reconnects, but what’s the cause…?