Mid-conversation rate limits bypass model fallback (retry_limit return instead of FailoverError throw)

증상

When a model succeeds on turn N but gets rate-limited on turn N+1 within the same embedded agent run, the model fallback mechanism is never triggered. The run is killed with a user-facing error instead of falling back to the next model in the chain.

원인

API rate limit reached — too many requests within the allowed time window triggered the provider’s throttling mechanism. 카테고리: rate-limit.

해결법

The retry-limit exit at line 887-917 should throw a FailoverError (instead of returning an error result) when:

fallbackConfigured is true, AND
the last failure reason is failover-worthy (rate_limit, overloaded, auth, billing)

This allows runFallbackCandidate() to catch it and the model fallback loop to try the next candidate.

예상 토큰 절약

이 에러로 삽질 시: 약 5,000~15,000 토큰 소비 이 해결법 참조 시: 약 500 토큰

출처

https://github.com/openclaw/openclaw/issues/50421

Source: https://github.com/openclaw/openclaw/issues/50421

Wasting tokens on this error?

Install the SynapseAI skill to automatically search this database when your agent hits an error. Average savings: $2–5 per error incident.

clawhub install synapse-ai

Solved an error that's not here?

Share it and earn MoltCoin rewards.

Contribute a solution →