Azure Cosmos DB for NoSQL returned an HTTP 408 request time out.
Troubleshooting steps
The following list contains known causes and solutions for request time out exceptions.
Check the service level agreement (SLA)
Check Azure Cosmos DB for NoSQL monitoring to see if the number of 408 exceptions violates the Azure Cosmos DB for NoSQL SLA.
Solution 1: It didn't violate the Azure Cosmos DB for NoSQL SLA
The application should handle this scenario and retry on these transient failures.
Solution 2: It did violate the Azure Cosmos DB for NoSQL SLA
Contact Azure Support.
Hot partition key
Azure Cosmos DB for NoSQL distributes the overall provisioned throughput evenly across physical partitions. When there's a hot partition, one or more logical partition keys on a physical partition are consuming all the physical partition's Request Units per second (RU/s). At the same time, the RU/s on other physical partitions are going unused. As a symptom, the total RU/s consumed are less than the overall provisioned RU/s at the database or container. You could still see throttling (429 errors) on the requests against the hot logical partition key. Use the Normalized RU Consumption metric to see if the workload is encountering a hot partition.
Solution
Choose a good partition key that evenly distributes request volume and storage. Learn how to change your partition key.
Related content
- Diagnose and troubleshoot issues when you use the Azure Cosmos DB for NoSQL .NET SDK.
- Learn about performance guidelines for .NET v3 and .NET v2.
- Diagnose and troubleshoot issues when you use the Azure Cosmos DB for NoSQL Java v4 SDK.
- Learn about performance guidelines for Java v4 SDK.