Skip to content

I am getting this issue : "Current OCU capacity limit has exceeded (Service: BedrockAgentRuntime) after my data source gets indexed. This issue occurs randomly, on re-indexing, it gives answers.

0

native_memory_throttle_exception: current OCU capacity limit has exceeded (Service: BedrockAgentRuntime, Status Code: 400, Request ID: af2e158a-193a-4a29-a882-75609bd4c99a) (SDK Attempt Count: 1)

But, the older KBs attached to another data source does not face these issue on the same region and from same account.

1 Answer
0

The error you're encountering is related to OpenSearch Compute Units (OCU) capacity limits in your Bedrock Knowledge Base. When you see "current OCU capacity limit has exceeded," it means your SearchOCU usage has reached the maximum configured limit for your account.

This issue can occur randomly because SearchOCU usage fluctuates based on several factors:

  1. The number of Knowledge Bases you have created (even if they're not actively being queried)
  2. The total size of your data (278 GiB across multiple Knowledge Bases can contribute to higher OCU usage)
  3. The complexity of search operations being performed

Even with relatively low request rates (under 100 requests per day), you can still hit OCU limits because OpenSearch Serverless maintains a minimum of 2 OCUs for search operations even when there's no activity, and scales based on factors beyond just request rate.

The reason your older Knowledge Bases aren't experiencing this issue might be related to their size, configuration, or the number of collections they use.

To address this issue, you can:

  1. Check your current SearchOCU limits in the AWS Management Console
  2. Request a quota increase by contacting AWS Support if needed
  3. Implement request rate limiting in your application
  4. Monitor your OCU usage with CloudWatch metrics
  5. Consider reviewing the number of Knowledge Bases you have, as the number of collections can influence SearchOCU consumption even with minimal query activity
  6. Implement retry logic with exponential backoff for failed requests

If the issue persists, you might want to analyze if there's been a significant increase in the number of searchable documents or fields across your indices, which could impact SearchOCU usage.
Sources
Monitoring OpenSearch Serverless events using Amazon EventBridge - Amazon OpenSearch Service
AWS OpenSearch SearchOCU keeps hitting the max limit | AWS re:Post
Troubleshooting Amazon Bedrock API Error Codes - Amazon Bedrock

answered a year ago

You are not logged in. Log in to post an answer.

A good answer clearly answers the question and provides constructive feedback and encourages professional growth in the question asker.