fix: keep bridge readiness stable during live state updates #33
Loading…
Reference in a new issue
No description provided.
Delete branch "fix/bridge-readiness"
Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?
Normal WLED state updates invalidated whole-Bridge readiness until each entity and availability ACK, causing 117/844 production probes to return 503 despite connected HA/MQTT. Readiness now retains confirmed initialization for the current HA generation during state/attribute writes. Every write and latest-revision recovery remains independent and ACK-bound; new entities, configuration changes, restore/reconnect and cleanup still invalidate initialization.
Three real HA/HTTP regressions cover held identical writes, rejection/stale ACK/recovery, new entities, configuration/reconnect/offline and successful probes with one async worker and held CPU capacity. Existing generation watchers reset on a new reconcile attempt. The previously failing resize fixture captures visible marker geometry atomically while preserving all assertions.
Validation: reproduced the original 503, targeted connector/HTTP/browser tests passed, then full just gate exit 0 (82 Rust tests, existing frontend/tooling suites, strict Clippy/deny/machete). Shared S3 cache verified without backend errors. JSON formats, automations, production resources and image namespace unchanged. Native CI/candidate acceptance required before merge; release/SBOM and primary kube GitOps follow.