Repository navigation
Conversation
Managed knowledge bases do not require a user-configured vector store, so the operator must accept their configuration and avoid self-managed index retries. Co-authored-by: Copilot <[email protected]>
boring-cyborg
Bot
added
area:providers
kind:documentation
provider:amazon
…-solution-pr-creation
Append the new knowledge_base_configuration parameter after all existing parameters so the operator's public signature stays compatible and add a regression test proving existing self-managed Dags still produce the same CreateKnowledgeBase call.
…-solution-pr-creation
|
@vincbeck — I've pushed updates addressing the backward-compatibility concern you raised, could you take another look? |
| if self.role_arn is None: | ||
| raise ValueError("`role_arn` must be specified to create a knowledge base.") |
There was a problem hiding this comment.
Why marking role_arn as optional parameter in its definition but not allowing it in the execution?
o-nikolas
left a comment
o-nikolas
left a comment
There was a problem hiding this comment.
Amazon provider pins botocore>=1.42.79 but support for this only arrived in botocore 1.43.32 so you will get some runtime errors from folks using the older version of botocore
| role_arn="arn:aws:iam::123456789012:role/BedrockExecutionRole", | ||
| knowledge_base_configuration={ | ||
| "type": "MANAGED", | ||
| "managedKnowledgeBaseConfiguration": {"embeddingModeType": "MANAGED"}, |
There was a problem hiding this comment.
embeddingModeType Is a typo right? should Model not Mode?
|
|
||
| .. code-block:: python | ||
|
|
||
| BedrockCreateKnowledgeBaseOperator( |
There was a problem hiding this comment.
Is it not worth updating the system test and using example include?
|
This pull request has been automatically marked as stale because it has not had recent activity. It will be closed in 5 days if no further activity occurs. Thank you for your contributions. |
github-actions
Bot
added
the
stale
Support creating Amazon Bedrock managed knowledge bases with
BedrockCreateKnowledgeBaseOperator.Managed knowledge bases do not require a user-configured vector store or vector index. The operator now accepts a custom knowledge-base configuration, omits storage configuration when it is not provided, preserves the existing self-managed VECTOR behavior, and skips vector-index retry handling for managed knowledge bases. Documentation and unit tests cover both direct configuration and
create_knowledge_base_kwargs.closes: #72592
Was generative AI tooling used to co-author this PR?
Generated-by: Copilot (GPT-5) following the guidelines