rkanumul commented on a change in pull request #6531:
URL: https://github.com/apache/incubator-pinot/pull/6531#discussion_r573550197



##########
File path: 
pinot-plugins/pinot-file-system/pinot-adls/src/main/java/org/apache/pinot/plugin/filesystem/ADLSGen2PinotFS.java
##########
@@ -106,24 +118,75 @@ public void init(PinotConfiguration config) {
     // TODO: consider to add the encryption of the following config
     String accessKey = config.getProperty(ACCESS_KEY);
     String fileSystemName = config.getProperty(FILE_SYSTEM_NAME);
+    String clientId = config.getProperty(CLIENT_ID);
+    String clientSecret = config.getProperty(CLIENT_SECRET);
+    String tenantId = config.getProperty(TENANT_ID);
 
     String dfsServiceEndpointUrl = HTTPS_URL_PREFIX + accountName + 
AZURE_STORAGE_DNS_SUFFIX;
     String blobServiceEndpointUrl = HTTPS_URL_PREFIX + accountName + 
AZURE_BLOB_DNS_SUFFIX;
 
-    StorageSharedKeyCredential sharedKeyCredential = new 
StorageSharedKeyCredential(accountName, accessKey);
+    DataLakeServiceClientBuilder dataLakeServiceClientBuilder = new 
DataLakeServiceClientBuilder().endpoint(dfsServiceEndpointUrl);
+    BlobServiceClientBuilder blobServiceClientBuilder = new 
BlobServiceClientBuilder().endpoint(blobServiceEndpointUrl);
+
+    if (accountName!= null && accessKey != null) {
+      LOGGER.info("Authenticating using the access key to the account.");
+      StorageSharedKeyCredential sharedKeyCredential = new 
StorageSharedKeyCredential(accountName, accessKey);
+      dataLakeServiceClientBuilder.credential(sharedKeyCredential);
+      blobServiceClientBuilder.credential(sharedKeyCredential);
+    } else if (clientId != null && clientSecret != null && tenantId != null) {
+      LOGGER.info("Authenticating using Azure Active Directory");
+      ClientSecretCredential clientSecretCredential = new 
ClientSecretCredentialBuilder()
+          .clientId(clientId)
+          .clientSecret(clientSecret)
+          .tenantId(tenantId)
+          .build();
+      dataLakeServiceClientBuilder.credential(clientSecretCredential);
+      blobServiceClientBuilder.credential(clientSecretCredential);
+    } else {
+      // Error out as at least one mode of auth info needed
+      throw new IllegalArgumentException("Expecting either (accountName, 
accessKey) or (clientId, clientSecret, tenantId)");
+    }
 
-    DataLakeServiceClient serviceClient = new 
DataLakeServiceClientBuilder().credential(sharedKeyCredential)
-        .endpoint(dfsServiceEndpointUrl)
-        .buildClient();
+    _blobServiceClient = blobServiceClientBuilder.buildClient();
+    DataLakeServiceClient serviceClient = 
dataLakeServiceClientBuilder.buildClient();
+    _fileSystemClient = getOrCreateClientWithFileSystem(serviceClient, 
fileSystemName);
 
-    _blobServiceClient =
-        new 
BlobServiceClientBuilder().credential(sharedKeyCredential).endpoint(blobServiceEndpointUrl).buildClient();
-    _fileSystemClient = serviceClient.getFileSystemClient(fileSystemName);
     LOGGER.info("ADLSGen2PinotFS is initialized (accountName={}, 
fileSystemName={}, dfsServiceEndpointUrl={}, "
             + "blobServiceEndpointUrl={}, enableChecksum={})", accountName, 
fileSystemName, dfsServiceEndpointUrl,
         blobServiceEndpointUrl, _enableChecksum);
   }
 
+  /**
+   * Returns the DataLakeFileSystemClient to the specified file system 
creating if it doesn't exist.
+   *
+   * @param serviceClient authenticated data lake service client to an account
+   * @param fileSystemName name of the file system (blob container)
+   * @return DataLakeFileSystemClient with the specified fileSystemName.
+   */
+  @VisibleForTesting
+  public DataLakeFileSystemClient 
getOrCreateClientWithFileSystem(DataLakeServiceClient serviceClient,

Review comment:
       There is some misunderstanding here, gen1 doesn't have the concept of 
blob store. I was equating gen2 blob with creating the directories in gen1. eg 
how we create pinot_segment/<cluster-name> automatically even though they are 
not pre created.
   I see what you are saying but don't quite agree with that. Sure, they can 
create/update it if they have particular need, but we provide a default.
   Don't agree with the "clean unintentionally created" part, that can happen 
to even directories with clustername or some other auto created path.




----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

For queries about this service, please contact Infrastructure at:
us...@infra.apache.org



---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscr...@pinot.apache.org
For additional commands, e-mail: commits-h...@pinot.apache.org

Reply via email to