Thanks — that's really useful, because your failure is actually one step earlier than mine, so we may be looking at two different problems. Does my device complete the TLS handshake with mgmt? Yes, fully. In my capture the OPTO1 opens TCP to mgmt.viessmann-platform.io:443, sends the ClientHello, gets back the same chain you describe — leaf *.viessmann-platform.io ← RapidSSL TLS RSA CA G1 ← DigiCert Global Root G2 — and then finishes the handshake and exchanges application data. So on my unit that DigiCert Global Root G2 chain is trusted; the handshake is not where mine dies. Where mine dies: right after that, my device resolves the broker production-7-10-rmq-broker.devices.viessmann-platform.io to 40.115.118.0 and then never sends a single packet to it — not even a SYN. The moment I present the same broker on a non‑.0 address (local DNS override → a TCP relay on 8883 to the real broker IP, TLS kept end‑to‑end), the MQTT/TLS session establishes immediately and ViCare comes back online. So my OPTO1 appears to refuse any broker IP ending in .0 (treating it as a network address). That's a Viessmann‑side bug: the broker simply landed on a .0 IP. Firmware version: I can't read it — it isn't exposed anywhere in ViCare while the device is offline, and I haven't found a remote way to query it. If you've found one, please share; otherwise I think only Viessmann support can tell us. On your DigiCert theory / your next step: your symptom — the device ACKs the certificate and then goes silent until the server closes ~30 s later — looks like the device rejecting the server certificate rather than a network problem. Two cheap things to check besides the power cycle: (1) the unit's clock/RTC — if the date is wildly wrong, a perfectly valid cert looks expired/not‑yet‑valid and the device aborts the handshake silently; (2) whether you see an actual TLS alert on the wire or just a server‑side FIN/RST with no alert. If after the power cycle your device does complete the mgmt handshake, watch what IP the broker resolves to for you — if it's also a .0, you'll very likely hit the same wall I did. To be clear, though: this is not a viable fix for me. The DNS override + proxy only runs because I'm keeping a laptop on for it, and that laptop isn't meant to be the household's DNS — I have no always‑on box to move it to. It's a diagnostic that proves the cause, not something I can leave running. What I actually expect is for Viessmann to ship a proper fix (move the broker off a .0 IP, or firmware that accepts it). Happy to compare captures if that helps.
... Mehr anzeigen