diff --git a/content/300-accelerate/650-troubleshoot.mdx b/content/300-accelerate/650-troubleshoot.mdx index 33fed6294f..e217a16dea 100644 --- a/content/300-accelerate/650-troubleshoot.mdx +++ b/content/300-accelerate/650-troubleshoot.mdx @@ -132,6 +132,36 @@ If the database is taking too long to respond to the connection request, Prisma **Suggested solution:** Verify that the database is active and reachable. If the database is in sleep mode, try to wake it up by sending a request to it using a direct database GUI tool or wake it up using the database's management console. +## [`P5011`](/orm/reference/error-reference#p5011-too-many-requests) (`TooManyRequests`) + +This error occurs when Prisma Accelerate detects a high volume of requests that surpasses allowable thresholds. It acts as a protective measure to safeguard both Prisma Accelerate and your underlying database from excessive load. + +### Possible causes for [`P5011`](/orm/reference/error-reference#p5011-too-many-requests) + +#### Aggressive retry loops + +If your application retries queries immediately or with minimal delay, especially after receiving certain errors, the rapid accumulation of requests can surpass the threshold. + +**Suggested solution:** +- Implement an exponential backoff strategy. Rather than retrying immediately or with a fixed delay, gradually increase the delay period after each failed attempt. +- This allows the system time to recover and reduces the likelihood of overwhelming Prisma Accelerate and your database. + +#### Sudden traffic spikes + +Unpredicted traffic surges (for example, during product launches, flash sales, or viral growth events) can cause the threshold to be met and result into `P5011`. + +**Suggested solution:** +- Consider proactive scaling strategies for both Prisma Accelerate and your database. +- Monitor traffic and resource usage. If you anticipate a surge, please contact [support](/platform/support) for capacity planning and potential configuration adjustments. + +#### Prolonged or planned high workloads + +Certain processes, such as bulk data imports, ETL operations, or extended CRON jobs, can generate continuous high query volume over time. + +**Suggested solution:** +- Use batching or chunking techniques to break large operations into smaller parts. +- Establish throttling or scheduling to distribute the load more evenly. + ## Other errors ### Error with MySQL (Aiven): "We were unable to process your request. Please refresh and try again."