chore(audit): track additional PowerSync log findings

[Artic0din]

Summary

Full audit of the supplied Home Assistant log from 16 July 2026 found additional PowerSync defects beyond the originally reported EV warnings and recorder flood.

Installed integration version: 2.12.857.
Log window: 03:02:09–06:10:30 Australia/Melbourne.

Existing related issues

• fix(ev): reconcile Tesla BLE pseudo vehicles with real VINs before dispatch #23 — Tesla BLE pseudo/real vehicle identity reconciliation and unsafe dispatch.
• fix(ev): treat ownership-blocked auto-schedule starts as skipped, not failed #24 — ownership-blocked Auto Schedule starts reported as failures.
• perf(power-sync): diagnose derived-template attribute amplification and residual sensor stalls #27 — residual PowerSync entity stalls after Home Assistant template cleanup.

New child scopes

• Redact Expo push credentials from logs.
• Prevent network-envelope startup reconciliation from blocking Home Assistant bootstrap.
• Recover site_info after transient failures.
• Coalesce concurrent site_info requests.
• Diagnose recurring LP infeasibility and stale-plan fallback.
• Back off recurring optional GloBird endpoint failures.
• Improve diagnostics for intermittent local Powerwall gateway timeouts.
• Rate-limit repeated PW3 placeholder-topology logging.

Excluded as standalone defects

The AEMO, Expo, and PowerSync API failures around 03:06–03:22 shared a DNS/network outage and later recovered. The single stale EV session cleanup and missing saved-reserve warning behaved fail-safe. Their relevant resilience implications are captured by the child issues rather than duplicated as provider-outage bugs.

Linear: https://linear.app/plaintextlabs/issue/PLA-153/choreaudit-track-additional-powersync-log-findings

---

Migration source: PLA-166
Linear status at cutover: Backlog
Cut over to GitHub on 2026-07-22.

[linear-code]

PLA-166