Errors
op_service_degraded
503 — Matter is operating in degraded mode; retry with backoff.
Last updated
Cause
Matter operates a load-shedding tier that returns op_service_degraded when platform-level signals (database connection saturation, queue backpressure, latency budget exhaustion) cross thresholds. Rather than letting a backend collapse and corrupt in-flight state, the platform sheds load by rejecting a fraction of incoming requests with a 503 and a Retry-After header.
This signal is platform-wide, not specific to your token or your resource. It typically clears within a minute or two as backends drain. The status page reflects ongoing incidents that are likely to produce a sustained 503 rate; transient one-off occurrences during traffic spikes are normal.
Fix
- Read the
Retry-Afterheader and wait at least that long before retrying. - Use exponential backoff with jitter for retries. Naive constant-delay retries amplify the load problem.
- Check the status page at
/runbooks/status-page-incidentsfor active incidents. Sustained 503 rates correspond to a posted incident. - If the workload is time-sensitive, surface the platform's reduced availability to the caller. The retry will eventually succeed; the platform does not skip the operation.
Related codes
op_provider_unavailable— a third-party provider, not Matter itself.op_rate_limit_exceeded— short-window per-token rate, not platform load.infra_internal_error— an unexpected platform error, not deliberate shedding.