Fix Confluence Cloud space sync failures with AWS Bedrock Knowledge Base

Summary

Confluence Cloud spaces may fail to sync with AWS Bedrock Knowledge Base when using a space filter if the friendly URL alias is used instead of the internal space key. This article explains how to identify the correct space key to ensure successful data ingestion. This applies to Atlassian Cloud.

Cause

The AWS Bedrock Confluence connector filters spaces by calling the Confluence REST API using the literal internal space key. In many Confluence Cloud instances, especially those created recently, the segment seen in the browser URL (the friendly URL alias) differs from the actual internal space key (often an auto-generated alphanumeric string). If the alias is used in the AWS Bedrock inclusion filter, the connector finds no matching resources and returns zero ingested files.

Solution

To resolve this issue, you must identify the internal space key for your Confluence space and update the inclusion filter in your AWS Bedrock data source configuration.

Step 1: Identify the internal space key

You can find the actual space key by calling the Confluence REST API. Replace <your-site> with your site name and <alias> with the segment from your space URL (e.g., /wiki/spaces/<alias>). Use the following API request:

In the JSON response, look for the key field. This value is the internal identifier required by the AWS Bedrock connector.

Step 2: Update the AWS Bedrock space filter

1. Log in to the AWS Management Console and navigate to Amazon Bedrock.

2. Go to Knowledge bases and select your specific knowledge base.

3. Select the Confluence data source and choose Edit.

4. In the Space filter (Include) section, remove the friendly URL alias.

5. Enter the internal space key identified in Step 1.

6. Save the data source configuration.

Step 3: Trigger a manual sync

1. On the data source details page, click Sync.

2. Monitor the ingestion job status.

3. Once the status is COMPLETE, verify that the number_of_resources_ingested is greater than zero.

4. Perform a test query in the Bedrock console to confirm content from the space is now searchable.

Recommendation: Use separate data sources

When you change a filter on an existing AWS Bedrock data source, the connector may treat previously indexed content from other spaces as stale and delete it. To prevent this, consider creating a separate data source for each Confluence space you wish to index.