YouTube API quota exceeded causes upload spam and series errors

Issue

The system experienced repeated failures when processing API quota limits, leading to inefficient resource usage and false errors. When API quotas were exceeded, the system would attempt to process subsequent API requests immediately instead of waiting for the quota to reset. This resulted in multiple consecutive quota-exceeded errors, excessive retries, and incorrect error messages regarding processing order, even though the operations were not truly out of sequence.

Culprit

The issue arose from a lack of global coordination for quota management. When a quota limit was reached, individual requests were re-queued with delays, but the main processing loop continued immediately, causing additional requests to be sent without waiting for the quota to reset. Additionally, the system did not differentiate between platform-specific quota limits, leading to race conditions when processing requests for different platforms.

Solution

Global Quota Backoff State

A global state mechanism was introduced to track quota backoff periods. This ensures that all API requests for a specific platform (e.g., YouTube) are paused until the quota limit resets, preventing repeated quota-exceeded errors.

Quota Error Handling

When a quota limit is exceeded, the system calculates the required backoff time and sets a global block for that platform. This ensures that no further requests are processed until the backoff period has elapsed, allowing the quota to reset.

Pre-Processing Checks

Before processing any requests, the system checks if a quota backoff period is active. If so, it skips the request and re-queues it with a priority that ensures it is processed only after the backoff period has ended.

Priority Queue Adjustment

The re-queue logic was adjusted to include the backoff expiry time as a priority component. This ensures that requests are processed in the correct order and that backoff periods are respected globally.

Platform Isolation

Non-quota-limited platforms continue to process requests normally during quota backoff periods, ensuring that the system remains functional and efficient for other operations.

The solution ensures that quota limits are respected globally, preventing repeated errors and improving system reliability.