Is it possible to add S3 datasource into OpenSearch using CDK?

[zentavr]

Hi,

The Amazon OpenSearch Service development guide says that it is possible to add S3 data source integration.

...like this:

How this could be done using CDK? I had not found any example to do so.

[Sagargupta16]

CDK doesn't have an L2 construct for OpenSearch direct query S3 datasources yet. You'll need to use the L1 CfnDataSource construct or AwsCustomResource.

Using AwsCustomResource

import { AwsCustomResource, AwsCustomResourcePolicy, PhysicalResourceId } from "aws-cdk-lib/custom-resources";

const dataSource = new AwsCustomResource(this, "OpenSearchS3DataSource", {
  onCreate: {
    service: "OpenSearchService",
    action: "createDataSource",
    parameters: {
      DomainName: domain.domainName,
      Name: "my-s3-datasource",
      DataSourceType: {
        S3GlueDataCatalog: {
          RoleArn: glueAccessRole.roleArn,
        },
      },
    },
    physicalResourceId: PhysicalResourceId.of("s3-datasource"),
  },
  onDelete: {
    service: "OpenSearchService",
    action: "deleteDataSource",
    parameters: {
      DomainName: domain.domainName,
      Name: "my-s3-datasource",
    },
  },
  policy: AwsCustomResourcePolicy.fromSdkCalls({ resources: AwsCustomResourcePolicy.ANY_RESOURCE }),
});

The S3 direct query feature works through Glue Data Catalog -- OpenSearch queries S3 data via a Glue catalog table, not directly. So you'll need:

1. A Glue database + table pointing to your S3 data
2. An IAM role with Glue and S3 permissions
3. The OpenSearch datasource linking to the Glue catalog

Reference: Creating direct query S3 datasources

There's also an open feature request for native L2 support -- worth upvoting if you find one, or opening a new one if not.