thelonglakelodge/reference/locks.md

Door Locks & Codes

Two Z-Wave (zwave_js) keypad locks: `lock.lodge_front_door_lock` (node 3) and `lock.bunkhouse_frontdoor_lock`. 8 shared code slots, managed from the "Lock Management" Lovelace view via helpers: `input_text.lock_slotN_user`, `input_text.lock_slotN_code`, `input_boolean.lock_slotN_enabled` (N = 1–8). Slot 1 = Fritz family, 2 = cleaner, 4 = carpenter, 5 = backup. All code writes go through `script.lock_sync_slot` (fields: `slot` 1–8), the single source of truth. Per slot it: resolves the code (slot's own code if `_enabled` on, else the slot-1 family code), REFUSES non-4-10-digit codes (persistent notification "slot N SKIPPED", nothing written), then per lock does a UserCode CC `get` (repopulates the Z-Wave value cache so a missing cached slot can't error) followed by `zwave_js.set_lock_usercode` — every zwave call `continue_on_error` so one lock/slot never blocks the rest.

• script.lock_sync_all_codes — warns (persistent + notify.mobile_app_donavan_s_iphone_16_pro) if a lock's *_node_status sensor isn't alive/asleep, then runs lock_sync_slot for slots 1–8.
• automation.lock_slot_N_sync_both_locks (×8) — trigger on the slot's _enabled OR _code helper change → lock_sync_slot for that slot.
• automation.lock_codes_full_sync_self_heal — full sync daily 04:00 and 10 min after HA start, so transient failures heal within a day.

Disabled slots are overwritten with the family code (design choice), so the family code exists in multiple lock slots.

Error `userCode for code slot N not found` from `zwave_js.set_lock_usercode` = Z-Wave JS value cache has no userCode value for that slot, NOT a lock or code problem. The sync script's pre-`get` prevents it; manual fix: `zwave_js.invoke_cc_api {device_id, command_class: "99", method_name: "get", parameters: [N]}` then re-run the sync.

Bunkhouse lock (Z-Wave node 8) intermittently returns ZW1405 - node failed to decode the message on commands (seen 2026-07-05 on unlock + usercode set). Sync tolerates it (continue_on_error + daily self-heal retries); if it persists, suspect S2/RF health on that lock.

Scripts/automations/dashboards/helper-definitions are mirrored to `dfritz/home-assistant` repo under `lodge/` (API export — the lodge HA has no SSH). Nightly Gitea Actions workflow `.gitea/workflows/backup-ha.yaml` IN THIS SKILL REPO (secrets `HA_TOKEN` + `GIT_TOKEN` on skills/thelonglakelodge; ops lacks rights to put secrets on dfritz/home-assistant) exports and pushes only on change; failure emails via action/notify-email. Manual run: the workflow's `workflow_dispatch`, or `HA_TOKEN=... lodge/backup.sh` in the mirror repo.