fix(snowflake-driver): Add support for IAM roles with IRSA for S3 export buckets

.github/workflows/drivers-tests.yml

@@ -204,6 +204,10 @@ jobs:
 
   tests:
     runs-on: ubuntu-24.04
+    permissions:
+      id-token: write   # Needed for OIDC+AWS
+      contents: read
+
     timeout-minutes: 30
     needs: [latest-tag-sha, build]
     if: (needs['latest-tag-sha'].outputs.sha != github.sha)
@@ -225,6 +229,7 @@ jobs:
         snowflake
         snowflake-encrypted-pk
         snowflake-export-bucket-s3
+        snowflake-export-bucket-s3-via-storage-integration-iam-roles
         snowflake-export-bucket-s3-prefix
         snowflake-export-bucket-azure
         snowflake-export-bucket-azure-prefix
@@ -259,6 +264,7 @@ jobs:
           - snowflake
           - snowflake-encrypted-pk
           - snowflake-export-bucket-s3
+          - snowflake-export-bucket-s3-via-storage-integration-iam-roles
           - snowflake-export-bucket-s3-prefix
           - snowflake-export-bucket-azure
           - snowflake-export-bucket-azure-prefix
@@ -338,6 +344,15 @@ jobs:
           gunzip image.tar.gz
           docker load -i image.tar
 
+      - name: Configure AWS credentials via IRSA
+        uses: aws-actions/configure-aws-credentials@v4
+        with:
+          role-to-assume: ${{ secrets.DRIVERS_TESTS_AWS_ROLE_ARN_FOR_SNOWFLAKE }}
+          aws-region: us-west-1
+          mask-aws-account-id: true
+        if: |
+          env.DRIVERS_TESTS_ATHENA_CUBEJS_AWS_KEY != '' && matrix.database == 'snowflake-export-bucket-s3-via-storage-integration-iam-roles'
+
       - name: Run tests
         uses: nick-fields/retry@v3
         # It's enough to test for any one secret because they are set all at once or not set all

docs/pages/product/configuration/data-sources/snowflake.mdx

@@ -133,15 +133,13 @@ Storage][google-cloud-storage] for export bucket functionality.
 
 <InfoBox>
 
-Ensure the AWS credentials are correctly configured in IAM to allow reads and
-writes to the export bucket in S3 if you are not using storage integration.
-If you are using storage integration then you still need to configure access keys
-for Cube Store to be able to read from the export bucket.
-It's possible to authenticate with IAM roles instead of access keys for Cube Store.
+Ensure proper IAM privileges are configured for S3 bucket reads and writes, using either 
+storage integration or user credentials for Snowflake and either IAM roles/IRSA or user 
+credentials for Cube Store, with mixed configurations supported.
 
 </InfoBox>
 
-Using IAM user credentials:
+Using IAM user credentials for both:
 
 ```dotenv
 CUBEJS_DB_EXPORT_BUCKET_TYPE=s3
@@ -151,8 +149,8 @@ CUBEJS_DB_EXPORT_BUCKET_AWS_SECRET=<AWS_SECRET>
 CUBEJS_DB_EXPORT_BUCKET_AWS_REGION=<AWS_REGION>
 ```
 
-[Using Storage Integration][snowflake-docs-aws-integration] to write to Export Bucket and 
-then Access Keys to read from Cube Store:
+Using a [Storage Integration][snowflake-docs-aws-integration] to write to export buckets and 
+user credentials to read from Cube Store:
 
 ```dotenv
 CUBEJS_DB_EXPORT_BUCKET_TYPE=s3
@@ -163,7 +161,8 @@ CUBEJS_DB_EXPORT_BUCKET_AWS_SECRET=<AWS_SECRET>
 CUBEJS_DB_EXPORT_BUCKET_AWS_REGION=<AWS_REGION>
 ```
 
-Using Storage Integration to write to export bocket and IAM role to read from Cube Store:
+Using a Storage Integration to write to export bucket and IAM role/IRSA to read from Cube Store:**
+
 ```dotenv
 CUBEJS_DB_EXPORT_BUCKET_TYPE=s3
 CUBEJS_DB_EXPORT_BUCKET=my.bucket.on.s3

packages/cubejs-snowflake-driver/package.json

@@ -25,6 +25,7 @@
     "lint:fix": "eslint --fix src/* --ext .ts"
   },
   "dependencies": {
+    "@aws-sdk/client-s3": "^3.726.0",
     "@cubejs-backend/base-driver": "1.3.77",
     "@cubejs-backend/shared": "1.3.77",
     "date-fns-timezone": "^0.1.4",

packages/cubejs-snowflake-driver/src/SnowflakeDriver.ts

@@ -23,6 +23,7 @@ import {
 import { formatToTimeZone } from 'date-fns-timezone';
 import fs from 'fs/promises';
 import crypto from 'crypto';
+import { S3ClientConfig } from '@aws-sdk/client-s3';
 import { HydrationMap, HydrationStream } from './HydrationStream';
 
 const SUPPORTED_BUCKET_TYPES = ['s3', 'gcs', 'azure'];
@@ -106,8 +107,8 @@ const SnowflakeToGenericType: Record<string, GenericDataBaseType> = {
 interface SnowflakeDriverExportAWS {
   bucketType: 's3',
   bucketName: string,
-  keyId: string,
-  secretKey: string,
+  keyId?: string,
+  secretKey?: string,
   region: string,
   integrationName?: string,
 }
@@ -328,14 +329,17 @@ export class SnowflakeDriver extends BaseDriver implements DriverInterface {
     if (bucketType === 's3') {
       // integrationName is optional for s3
       const integrationName = getEnv('dbExportIntegration', { dataSource });
+      // keyId and secretKey are optional for s3 if IAM role is used
+      const keyId = getEnv('dbExportBucketAwsKey', { dataSource });
+      const secretKey = getEnv('dbExportBucketAwsSecret', { dataSource });
 
       return {
         bucketType,
         bucketName: getEnv('dbExportBucket', { dataSource }),
-        keyId: getEnv('dbExportBucketAwsKey', { dataSource }),
-        secretKey: getEnv('dbExportBucketAwsSecret', { dataSource }),
         region: getEnv('dbExportBucketAwsRegion', { dataSource }),
         ...(integrationName !== undefined && { integrationName }),
+        ...(keyId !== undefined && { keyId }),
+        ...(secretKey !== undefined && { secretKey }),
       };
     }
 
@@ -387,6 +391,20 @@ export class SnowflakeDriver extends BaseDriver implements DriverInterface {
     );
   }
 
+  private getRequiredExportBucketKeys(
+    exportBucket: SnowflakeDriverExportBucket,
+    emptyKeys: string[]
+  ): string[] {
+    if (exportBucket.bucketType === 's3') {
+      const s3Config = exportBucket as SnowflakeDriverExportAWS;
+      if (s3Config.integrationName) {
+        return emptyKeys.filter(key => key !== 'keyId' && key !== 'secretKey');
+      }
+    }
+
+    return emptyKeys;
+  }
+
   protected getExportBucket(
     dataSource: string,
   ): SnowflakeDriverExportBucket | undefined {
@@ -402,9 +420,11 @@ export class SnowflakeDriver extends BaseDriver implements DriverInterface {
 
       const emptyKeys = Object.keys(exportBucket)
         .filter((key: string) => exportBucket[<keyof SnowflakeDriverExportBucket>key] === undefined);
-      if (emptyKeys.length) {
+      const keysToValidate = this.getRequiredExportBucketKeys(exportBucket, emptyKeys);
+
+      if (keysToValidate.length) {
         throw new Error(
-          `Unsupported configuration exportBucket, some configuration keys are empty: ${emptyKeys.join(',')}`
+          `Unsupported configuration exportBucket, some configuration keys are empty: ${keysToValidate.join(',')}`
         );
       }
 
@@ -731,7 +751,7 @@ export class SnowflakeDriver extends BaseDriver implements DriverInterface {
       // Storage integration export flow takes precedence over direct auth if it is defined
       if (conf.integrationName) {
         optionsToExport.STORAGE_INTEGRATION = conf.integrationName;
-      } else {
+      } else if (conf.keyId && conf.secretKey) {
         optionsToExport.CREDENTIALS = `(AWS_KEY_ID = '${conf.keyId}' AWS_SECRET_KEY = '${conf.secretKey}')`;
       }
     } else if (bucketType === 'gcs') {
@@ -771,14 +791,18 @@ export class SnowflakeDriver extends BaseDriver implements DriverInterface {
       const { bucketName, path } = this.parseBucketUrl(this.config.exportBucket!.bucketName);
       const exportPrefix = path ? `${path}/${tableName}` : tableName;
 
+      const s3Config: S3ClientConfig = { region };
+      if (keyId && secretKey) {
+        // If access key and secret are provided, use them as credentials
+        // Otherwise, let the SDK use the default credential chain (IRSA, instance profile, etc.)
+        s3Config.credentials = {
+          accessKeyId: keyId,
+          secretAccessKey: secretKey,
+        };
+      }
+
       return this.extractUnloadedFilesFromS3(
-        {
-          credentials: {
-            accessKeyId: keyId,
-            secretAccessKey: secretKey,
-          },
-          region,
-        },
+        s3Config,
         bucketName,
         exportPrefix,
       );

packages/cubejs-testing-drivers/fixtures/snowflake.json

@@ -22,6 +22,21 @@
         }
       }
     },
+    "export-bucket-s3-via-storage-integration-iam-roles": {
+      "cube": {
+        "environment": {
+          "CUBEJS_DB_EXPORT_BUCKET_TYPE": "s3",
+          "CUBEJS_DB_EXPORT_BUCKET": "snowflake-drivers-tests-preaggs",
+          "CUBEJS_DB_EXPORT_BUCKET_AWS_REGION": "us-west-1",
+          "CUBEJS_DB_EXPORT_INTEGRATION": "DRIVERS_TESTS_PREAGGS_S3",
+          "AWS_REGION": "us-west-1",
+          "AWS_ACCESS_KEY_ID": "${AWS_ACCESS_KEY_ID}",
+          "AWS_SECRET_ACCESS_KEY": "${AWS_SECRET_ACCESS_KEY}",
+          "AWS_SESSION_TOKEN": "${AWS_SESSION_TOKEN}",
+          "AWS_DEFAULT_REGION": "${AWS_DEFAULT_REGION}"
+        }
+      }
+    },
     "export-bucket-azure": {
       "cube": {
         "environment": {

packages/cubejs-testing-drivers/package.json

@@ -49,6 +49,7 @@
     "snowflake-full": "yarn test-driver -i dist/test/snowflake-full.test.js",
     "snowflake-encrypted-pk-full": "yarn test-driver -i dist/test/snowflake-encrypted-pk-full.test.js",
     "snowflake-export-bucket-s3-full": "yarn test-driver -i dist/test/snowflake-export-bucket-s3-full.test.js",
+    "snowflake-export-bucket-s3-via-storage-integration-iam-roles-full": "yarn test-driver -i dist/test/snowflake-export-bucket-s3-via-storage-integration-iam-roles-full.test.js",
     "snowflake-export-bucket-s3-prefix-full": "yarn test-driver -i dist/test/snowflake-export-bucket-s3-prefix-full.test.js",
     "snowflake-export-bucket-azure-full": "yarn test-driver -i dist/test/snowflake-export-bucket-azure-full.test.js",
     "snowflake-export-bucket-azure-prefix-full": "yarn test-driver -i dist/test/snowflake-export-bucket-azure-prefix-full.test.js",
@@ -59,7 +60,7 @@
     "redshift-core": "yarn test-driver -i dist/test/redshift-core.test.js",
     "redshift-full": "yarn test-driver -i dist/test/redshift-full.test.js",
     "redshift-export-bucket-s3-full": "yarn test-driver -i dist/test/redshift-export-bucket-s3-full.test.js",
-    "update-all-snapshots-local": "yarn run athena-export-bucket-s3-full --mode=local -u; yarn run bigquery-export-bucket-gcs-full --mode=local -u; yarn run clickhouse-full --mode=local -u; yarn run clickhouse-export-bucket-s3-full --mode=local -u; yarn run clickhouse-export-bucket-s3-prefix-full --mode=local -u; yarn run databricks-jdbc-export-bucket-azure-full --mode=local -u; yarn run databricks-jdbc-export-bucket-azure-prefix-full --mode=local -u; yarn run databricks-jdbc-export-bucket-gcs-full --mode=local -u; yarn run databricks-jdbc-export-bucket-gcs-prefix-full --mode=local -u; yarn run databricks-jdbc-export-bucket-s3-full --mode=local -u; yarn run databricks-jdbc-export-bucket-s3-prefix-full --mode=local -u; yarn run databricks-jdbc-full --mode=local -u; yarn run mssql-full --mode=local -u; yarn run mysql-full --mode=local -u; yarn run postgres-full --mode=local -u; yarn run redshift-export-bucket-s3-full --mode=local -u; yarn run redshift-full --mode=local -u; yarn run snowflake-encrypted-pk-full --mode=local -u; yarn run snowflake-export-bucket-azure-full --mode=local -u; yarn run snowflake-export-bucket-azure-prefix-full --mode=local -u; yarn run snowflake-export-bucket-azure-via-storage-integration-full --mode=local -u; yarn run snowflake-export-bucket-gcs-full --mode=local -u; yarn run snowflake-export-bucket-gcs-prefix-full --mode=local -u; yarn run snowflake-export-bucket-s3-full --mode=local -u; yarn run snowflake-export-bucket-s3-prefix-full --mode=local -u; yarn run snowflake-export-bucket-azure-prefix-full --mode=local -u; yarn run snowflake-export-bucket-azure-full --mode=local -u; yarn run snowflake-full --mode=local -u",
+    "update-all-snapshots-local": "yarn run athena-export-bucket-s3-full --mode=local -u; yarn run bigquery-export-bucket-gcs-full --mode=local -u; yarn run clickhouse-full --mode=local -u; yarn run clickhouse-export-bucket-s3-full --mode=local -u; yarn run clickhouse-export-bucket-s3-prefix-full --mode=local -u; yarn run databricks-jdbc-export-bucket-azure-full --mode=local -u; yarn run databricks-jdbc-export-bucket-azure-prefix-full --mode=local -u; yarn run databricks-jdbc-export-bucket-gcs-full --mode=local -u; yarn run databricks-jdbc-export-bucket-gcs-prefix-full --mode=local -u; yarn run databricks-jdbc-export-bucket-s3-full --mode=local -u; yarn run databricks-jdbc-export-bucket-s3-prefix-full --mode=local -u; yarn run databricks-jdbc-full --mode=local -u; yarn run mssql-full --mode=local -u; yarn run mysql-full --mode=local -u; yarn run postgres-full --mode=local -u; yarn run redshift-export-bucket-s3-full --mode=local -u; yarn run redshift-full --mode=local -u; yarn run snowflake-encrypted-pk-full --mode=local -u; yarn run snowflake-export-bucket-azure-full --mode=local -u; yarn run snowflake-export-bucket-azure-prefix-full --mode=local -u; yarn run snowflake-export-bucket-azure-via-storage-integration-full --mode=local -u; yarn run snowflake-export-bucket-gcs-full --mode=local -u; yarn run snowflake-export-bucket-gcs-prefix-full --mode=local -u; yarn run snowflake-export-bucket-s3-full --mode=local -u; yarn run snowflake-export-bucket-s3-via-storage-integration-iam-roles-full --mode=local -u; yarn run snowflake-export-bucket-s3-prefix-full --mode=local -u; yarn run snowflake-export-bucket-azure-prefix-full --mode=local -u; yarn run snowflake-export-bucket-azure-full --mode=local -u; yarn run snowflake-full --mode=local -u",
     "tst": "clear && yarn tsc && yarn bigquery-core"
   },
   "files": [