Cloud Sync Resilience

Telemetry upload to POST /api/v1/ledger/events uses bounded timeouts, retries with jitter, and respects HTTP 429.

Agent hint

Use for sync_error_count, events_dropped, sync_max_retries — not provider LLM retry.

In scope

  • Cloud sync retry behavior
  • Health fields for sync failures
  • sync_timeout / syncMaxRetries config

Out of scope

  • Provider-call retry/fallback — see /docs/reliability
  • Backend rate limit numbers — see Rate Limits
This is cloud sync resilience (SDK → Aurex backend). Provider-call resilience (OpenAI → fallback model) is documented under Reliability & Loss Recovery.
PYTHON
health = auditor.get_health()
# {
#   "status": "healthy" | "degraded",
#   "queue_size": 0,
#   "events_dropped": 0,
#   "sync_error_count": 0,
#   "storage_mode": "file",
#   "ledger_path": "/tmp/.aurex/ledger.12345.jsonl"
# }

See also