openclaw - 💡(How to fix) Fix Make `PRIMARY_MODEL_PREWARM_TIMEOUT_MS` configurable via env or config (follow-up to #74776) [2 comments, 2 participants]

TL;DR

Increase the startup model warmup timeout by configuring gateway.warmupTimeoutMs or setting the OPENCLAW_PRIMARY_MODEL_PREWARM_TIMEOUT_MS environment variable to a value greater than 5000ms.

Guidance

• Identify the current timeout value and consider increasing it based on the specific runtime environment, such as gemini-cli or pi, to accommodate slower auth refresh times.
• Review the proposed fix and consider implementing it to make the timeout configurable via gateway.warmupTimeoutMs or OPENCLAW_PRIMARY_MODEL_PREWARM_TIMEOUT_MS.
• To verify the fix, restart the gateway with the updated configuration and check for the absence of the "startup model warmup timed out" warning.
• Consider testing with different timeout values to determine the optimal setting for the specific use case.

Example

// Example configuration update
const config = {
  // ...
  gateway: {
    warmupTimeoutMs: 15000, // increased timeout value
  },
};

Notes

The proposed fix aims to address the cosmetic issue of the startup model warmup timeout warning, but it may not resolve the underlying issues of follow-up failures (fetch-timeout, model-call hangs). Further investigation may be necessary to address these related problems.

Recommendation

Apply the proposed fix by configuring gateway.warmupTimeoutMs or setting the OPENCLAW_PRIMARY_MODEL_PREWARM_TIMEOUT_MS environment variable to increase the startup model warmup timeout, as this will help reduce the noise warnings and potentially improve the overall stability of the gateway.