Method set_quota_exceeded

config/global_state.py:100–142

Sets the global quota exceeded flag and logs a critical warning. Also determines the error type based on the message for adaptive cooldowns. Optionally accepts a model_id to flag specific model exhaustion.

(cls, message: str = "", model_id: Optional[str] = None)

    @classmethod
    def set_quota_exceeded(cls, message: str = "", model_id: Optional[str] = None):
        """
        Sets the global quota exceeded flag and logs a critical warning.
        Also determines the error type based on the message for adaptive cooldowns.
        Optionally accepts a model_id to flag specific model exhaustion.
        """
        if not cls.IS_QUOTA_EXCEEDED:
            cls.IS_QUOTA_EXCEEDED = True
            cls.QUOTA_EXCEEDED_TIMESTAMP = time.time()
            cls.QUOTA_EXCEEDED_EVENT.set()

            # Determine error type
            safe_message = message if message else ""
            msg_lower = safe_message.lower()
            if (
                "429" in msg_lower
                or "rate limit" in msg_lower
                or "resource has been exhausted" in msg_lower
            ):
                # API "RESOURCE_EXHAUSTED" usually means 429/quota shared behavior,
                # but "rate limit" specifically implies a temporary 429.
                # However, Gemini "Resource has been exhausted" is often a harder limit.
                # Let's verify standard Gemini strings:
                # "429: Too Many Requests" -> Rate Limit
                # "429: Resource has been exhausted" -> Quota
                if "too many requests" in msg_lower:
                    cls.last_error_type = "RATE_LIMIT"
                else:
                    cls.last_error_type = "QUOTA_EXCEEDED"
            elif "quota" in msg_lower:
                cls.last_error_type = "QUOTA_EXCEEDED"
            else:
                # Default fallback if unknown
                cls.last_error_type = "QUOTA_EXCEEDED"

            # [FIX] If model_id is provided, immediately mark it as exhausted so rotation logic knows
            if model_id and cls.last_error_type == "QUOTA_EXCEEDED":
                cls.current_profile_exhausted_models.add(model_id.lower())
                logger.warning(f"⛔ Identified specific model exhaustion: {model_id}")

            logger.critical(
                f"⛔ GLOBAL ALERT: Quota Exceeded! Type: {cls.last_error_type} (Event Signal Sent)"
            )

    @classmethod
    def reset_quota_status(cls):