Update: False alarm (sort of)
Good news: I come bearing data and a mild dose of humble pie.
After posting my dire warning about 1.0.48, I dug into the Home Assistant database to see what actually happened —
and it turns out the Levven API is not dead. It was just… rebooting.
The “broken” behavior I observed (no punt!\n greeting, connection reset on the magic header) was apparently the spa mid-firmware-flash, which is not exactly its most communicative state. The local API came back on its own about 16 minutes after the update completed.
That said, 1.0.48 is moodier than 1.0.43. The connection dropped more frequently in the days following the update — sometimes every few hours, occasionally in rapid bursts — before settling down and becoming completely stable again by day 5. The coordinator’s auto-reconnect handled it each time, but entities would flicker to unknown in the dashboard during the gaps.
I’ve since tightened up the custom component to deal with this more gracefully:
- The greeting (punt!\n) is now treated as optional — if it doesn’t arrive within 3 seconds, the coordinator proceeds anyway instead of bailing
- Connection resets now trigger a 3-second reconnect instead of 10
- Short drops (< 60 seconds) no longer flash unknown in the UI
So: if you updated to 1.0.48 and the integration stopped working, give it 20–30 minutes and a power cycle if needed. It most likely comes back. If you’re seeing frequent reconnections, the coordinator fix above should help.
Sorry for the alarm — turns out the patient just needed a nap. 