Summary

Had Linkind/AiDot Matter bulbs randomly go Unavailable in HA — real CASE session timeouts in the matter-server logs, not a Wi-Fi issue (device still showed up on mDNS, zero deauth events on my router the whole time). HA's own "reload Matter integration" did nothing. Toggling the bulb from the AiDot manufacturer app — a separate Matter fabric — fixed it in under a minute, every time, even after a 2.5-hour outage. These bulbs were also commissioned to 3 fabrics at once (HA, Alexa via Matterbridge, AiDot app), which lines up with a plausible-but-unconfirmed theory that it's a multi-fabric session-slot issue on the bulb's side. Did a deep dive through GitHub issues, HA forums, and vendor docs — the general "CASE session dies while mDNS still works" bug is well known and unsolved, but this specific fix doesn't seem to be written up anywhere else. Full write-up: https://claude.ai/artifact/NnsoTie8vzQg2nnFkdev2h Anyone else running multi-fabric Matter bulbs hit this?