Add EMR client to spark connector (opensearch-project#1790)

* Create Spark Connector Signed-off-by: Vamsi Manohar <[email protected]> * Add spark client and engine Signed-off-by: Rupal Mahajan <[email protected]> * Remove vars Signed-off-by: Rupal Mahajan <[email protected]> * Spark connector draft Signed-off-by: Rupal Mahajan <[email protected]> * nit Signed-off-by: Rupal Mahajan <[email protected]> * Fix checkstyle errors Signed-off-by: Rupal Mahajan <[email protected]> * nit Signed-off-by: Rupal Mahajan <[email protected]> * Fix license header Signed-off-by: Rupal Mahajan <[email protected]> * Add spark storage test Signed-off-by: Rupal Mahajan <[email protected]> * Update comments Signed-off-by: Rupal Mahajan <[email protected]> * Fix checkstyle in comments Signed-off-by: Rupal Mahajan <[email protected]> * Update tests Signed-off-by: Rupal Mahajan <[email protected]> * Add emr client Signed-off-by: Rupal Mahajan <[email protected]> * Set default values for flint args Signed-off-by: Rupal Mahajan <[email protected]> * Validate emr auth type Signed-off-by: Rupal Mahajan <[email protected]> * Add default constants for flint Signed-off-by: Rupal Mahajan <[email protected]> * Update unit tests Signed-off-by: Rupal Mahajan <[email protected]> * Address PR comments Signed-off-by: Rupal Mahajan <[email protected]> * tests draft Signed-off-by: Rupal Mahajan <[email protected]> * Refactor class name Signed-off-by: Rupal Mahajan <[email protected]> * Rename classes and update tests Signed-off-by: Rupal Mahajan <[email protected]> * Update scan operator test Signed-off-by: Rupal Mahajan <[email protected]> * Address PR comment Signed-off-by: Rupal Mahajan <[email protected]> * Fix Connection pool shut down issue Signed-off-by: Rupal Mahajan <[email protected]> * Update emr unit tests Signed-off-by: Rupal Mahajan <[email protected]> * Update doc and tests Signed-off-by: Rupal Mahajan <[email protected]> * nit Signed-off-by: Rupal Mahajan <[email protected]> * Update EMR clinet impl tests Signed-off-by: Rupal Mahajan <[email protected]> * Address PR comments Signed-off-by: Rupal Mahajan <[email protected]> * Make spark & flint jars configurable Signed-off-by: Rupal Mahajan <[email protected]> * Address comments Signed-off-by: Rupal Mahajan <[email protected]> * Add spark application id in logs Signed-off-by: Rupal Mahajan <[email protected]> * nit Signed-off-by: Rupal Mahajan <[email protected]> * Delete result when not required Signed-off-by: Rupal Mahajan <[email protected]> --------- Signed-off-by: Vamsi Manohar <[email protected]> Signed-off-by: Rupal Mahajan <[email protected]> Co-authored-by: Vamsi Manohar <[email protected]>
Bit-Quill · Jul 11, 2023 · 392a720 · 392a720
1 parent ab375e8
commit 392a720
Show file tree

Hide file tree

Showing 30 changed files with 1,559 additions and 44 deletions.
diff --git a/DEVELOPER_GUIDE.rst b/DEVELOPER_GUIDE.rst
@@ -147,6 +147,7 @@ The plugin codebase is in standard layout of Gradle project::
    ├── plugin
    ├── protocol
    ├── ppl
+   ├── spark
    ├── sql
    ├── sql-cli
    ├── sql-jdbc
@@ -161,6 +162,7 @@ Here are sub-folders (Gradle modules) for plugin source code:
 - ``core``: core query engine.
 - ``opensearch``: OpenSearch storage engine.
 - ``prometheus``: Prometheus storage engine.
+- ``spark`` : Spark storage engine
 - ``protocol``: request/response protocol formatter.
 - ``common``: common util code.
 - ``integ-test``: integration and comparison test.

diff --git a/docs/user/ppl/admin/spark_connector.rst b/docs/user/ppl/admin/spark_connector.rst
@@ -0,0 +1,92 @@
+.. highlight:: sh
+
+====================
+Spark Connector
+====================
+
+.. rubric:: Table of contents
+
+.. contents::
+   :local:
+   :depth: 1
+
+
+Introduction
+============
+
+This page covers spark connector properties for dataSource configuration
+and the nuances associated with spark connector.
+
+
+Spark Connector Properties in DataSource Configuration
+========================================================
+Spark Connector Properties.
+
+* ``spark.connector`` [Required].
+    * This parameters provides the spark client information for connection.
+* ``spark.sql.application`` [Optional].
+    * This parameters provides the spark sql application jar. Default value is ``s3://spark-datasource/sql-job.jar``.
+* ``emr.cluster`` [Required].
+    * This parameters provides the emr cluster id information.
+* ``emr.auth.type`` [Required]
+    * This parameters provides the authentication type information.
+    * Spark emr connector currently supports ``awssigv4`` authentication mechanism and following parameters are required.
+        * ``emr.auth.region``, ``emr.auth.access_key`` and ``emr.auth.secret_key``
+* ``spark.datasource.flint.*`` [Optional]
+    * This parameters provides the Opensearch domain host information for flint integration.
+    * ``spark.datasource.flint.integration`` [Optional]
+        * Default value for integration jar is ``s3://spark-datasource/flint-spark-integration-assembly-0.1.0-SNAPSHOT.jar``.
+    * ``spark.datasource.flint.host`` [Optional]
+        * Default value for host is ``localhost``.
+    * ``spark.datasource.flint.port`` [Optional]
+        * Default value for port is ``9200``.
+    * ``spark.datasource.flint.scheme`` [Optional]
+        * Default value for scheme is ``http``.
+    * ``spark.datasource.flint.auth`` [Optional]
+        * Default value for auth is ``false``.
+    * ``spark.datasource.flint.region`` [Optional]
+        * Default value for auth is ``us-west-2``.
+
+Example spark dataSource configuration
+========================================
+
+AWSSigV4 Auth::
+
+    [{
+        "name" : "my_spark",
+        "connector": "spark",
+        "properties" : {
+            "spark.connector": "emr",
+            "emr.cluster" : "{{clusterId}}",
+            "emr.auth.type" : "awssigv4",
+            "emr.auth.region" : "us-east-1",
+            "emr.auth.access_key" : "{{accessKey}}"
+            "emr.auth.secret_key" : "{{secretKey}}"
+            "spark.datasource.flint.host" : "{{opensearchHost}}",
+            "spark.datasource.flint.port" : "{{opensearchPort}}",
+            "spark.datasource.flint.scheme" : "{{opensearchScheme}}",
+            "spark.datasource.flint.auth" : "{{opensearchAuth}}",
+            "spark.datasource.flint.region" : "{{opensearchRegion}}",
+        }
+    }]
+
+
+Spark SQL Support
+==================
+
+`sql` Function
+----------------------------
+Spark connector offers `sql` function. This function can be used to run spark sql query.
+The function takes spark sql query as input. Argument should be either passed by name or positionArguments should be either passed by name or position.
+`source=my_spark.sql('select 1')`
+or
+`source=my_spark.sql(query='select 1')`
+Example::
+
+    > source=my_spark.sql('select 1')
+    +---+
+    | 1 |
+    |---+
+    | 1 |
+    +---+
+
diff --git a/spark/build.gradle b/spark/build.gradle
@@ -19,10 +19,12 @@ dependencies {
 
     implementation group: 'org.opensearch', name: 'opensearch', version: "${opensearch_version}"
     implementation group: 'org.json', name: 'json', version: '20230227'
+    implementation group: 'com.amazonaws', name: 'aws-java-sdk-emr', version: '1.12.1'
 
     testImplementation('org.junit.jupiter:junit-jupiter:5.6.2')
-    testImplementation group: 'org.mockito', name: 'mockito-core', version: '3.12.4'
-    testImplementation group: 'org.mockito', name: 'mockito-junit-jupiter', version: '3.12.4'
+    testImplementation group: 'org.mockito', name: 'mockito-core', version: '5.2.0'
+    testImplementation group: 'org.mockito', name: 'mockito-junit-jupiter', version: '5.2.0'
+    testImplementation 'junit:junit:4.13.1'
 }
 
 test {

diff --git a/spark/src/main/java/org/opensearch/sql/spark/client/EmrClientImpl.java b/spark/src/main/java/org/opensearch/sql/spark/client/EmrClientImpl.java
@@ -0,0 +1,121 @@
+/*
+ * Copyright OpenSearch Contributors
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+package org.opensearch.sql.spark.client;
+
+import static org.opensearch.sql.spark.data.constants.SparkConstants.SPARK_INDEX_NAME;
+import static org.opensearch.sql.spark.data.constants.SparkConstants.SPARK_SQL_APPLICATION_JAR;
+
+import com.amazonaws.services.elasticmapreduce.AmazonElasticMapReduce;
+import com.amazonaws.services.elasticmapreduce.model.ActionOnFailure;
+import com.amazonaws.services.elasticmapreduce.model.AddJobFlowStepsRequest;
+import com.amazonaws.services.elasticmapreduce.model.AddJobFlowStepsResult;
+import com.amazonaws.services.elasticmapreduce.model.DescribeStepRequest;
+import com.amazonaws.services.elasticmapreduce.model.HadoopJarStepConfig;
+import com.amazonaws.services.elasticmapreduce.model.StepConfig;
+import com.amazonaws.services.elasticmapreduce.model.StepStatus;
+import com.google.common.annotations.VisibleForTesting;
+import java.io.IOException;
+import lombok.SneakyThrows;
+import org.apache.logging.log4j.LogManager;
+import org.apache.logging.log4j.Logger;
+import org.json.JSONObject;
+import org.opensearch.sql.spark.helper.FlintHelper;
+import org.opensearch.sql.spark.response.SparkResponse;
+
+public class EmrClientImpl implements SparkClient {
+  private final AmazonElasticMapReduce emr;
+  private final String emrCluster;
+  private final FlintHelper flint;
+  private final String sparkApplicationJar;
+  private static final Logger logger = LogManager.getLogger(EmrClientImpl.class);
+  private SparkResponse sparkResponse;
+
+  /**
+   * Constructor for EMR Client Implementation.
+   *
+   * @param emr     EMR helper
+   * @param flint   Opensearch args for flint integration jar
+   * @param sparkResponse Response object to help with retrieving results from Opensearch index
+   */
+  public EmrClientImpl(AmazonElasticMapReduce emr, String emrCluster, FlintHelper flint,
+                       SparkResponse sparkResponse, String sparkApplicationJar) {
+    this.emr = emr;
+    this.emrCluster = emrCluster;
+    this.flint = flint;
+    this.sparkResponse = sparkResponse;
+    this.sparkApplicationJar =
+        sparkApplicationJar == null ? SPARK_SQL_APPLICATION_JAR : sparkApplicationJar;
+  }
+
+  @Override
+  public JSONObject sql(String query) throws IOException {
+    runEmrApplication(query);
+    return sparkResponse.getResultFromOpensearchIndex();
+  }
+
+  @VisibleForTesting
+  void runEmrApplication(String query) {
+
+    HadoopJarStepConfig stepConfig = new HadoopJarStepConfig()
+        .withJar("command-runner.jar")
+        .withArgs("spark-submit",
+            "--class","org.opensearch.sql.SQLJob",
+            "--jars",
+            flint.getFlintIntegrationJar(),
+            sparkApplicationJar,
+            query,
+            SPARK_INDEX_NAME,
+            flint.getFlintHost(),
+            flint.getFlintPort(),
+            flint.getFlintScheme(),
+            flint.getFlintAuth(),
+            flint.getFlintRegion()
+        );
+
+    StepConfig emrstep = new StepConfig()
+        .withName("Spark Application")
+        .withActionOnFailure(ActionOnFailure.CONTINUE)
+        .withHadoopJarStep(stepConfig);
+
+    AddJobFlowStepsRequest request = new AddJobFlowStepsRequest()
+        .withJobFlowId(emrCluster)
+        .withSteps(emrstep);
+
+    AddJobFlowStepsResult result = emr.addJobFlowSteps(request);
+    logger.info("EMR step ID: " + result.getStepIds());
+
+    String stepId = result.getStepIds().get(0);
+    DescribeStepRequest stepRequest = new DescribeStepRequest()
+        .withClusterId(emrCluster)
+        .withStepId(stepId);
+
+    waitForStepExecution(stepRequest);
+    sparkResponse.setValue(stepId);
+  }
+
+  @SneakyThrows
+  private void waitForStepExecution(DescribeStepRequest stepRequest) {
+    // Wait for the step to complete
+    boolean completed = false;
+    while (!completed) {
+      // Get the step status
+      StepStatus statusDetail = emr.describeStep(stepRequest).getStep().getStatus();
+      // Check if the step has completed
+      if (statusDetail.getState().equals("COMPLETED")) {
+        completed = true;
+        logger.info("EMR step completed successfully.");
+      } else if (statusDetail.getState().equals("FAILED")
+          || statusDetail.getState().equals("CANCELLED")) {
+        logger.error("EMR step failed or cancelled.");
+        throw new RuntimeException("Spark SQL application failed.");
+      } else {
+        // Sleep for some time before checking the status again
+        Thread.sleep(2500);
+      }
+    }
+  }
+
+}
diff --git a/spark/src/main/java/org/opensearch/sql/spark/data/constants/SparkConstants.java b/spark/src/main/java/org/opensearch/sql/spark/data/constants/SparkConstants.java
@@ -0,0 +1,20 @@
+/*
+ * Copyright OpenSearch Contributors
+ * SPDX-License-Identifier: Apache-2.0
+ */
+
+package org.opensearch.sql.spark.data.constants;
+
+public class SparkConstants {
+  public static final String EMR = "emr";
+  public static final String STEP_ID_FIELD = "stepId.keyword";
+  public static final String SPARK_SQL_APPLICATION_JAR = "s3://spark-datasource/sql-job.jar";
+  public static final String SPARK_INDEX_NAME = ".query_execution_result";
+  public static final String FLINT_INTEGRATION_JAR =
+      "s3://spark-datasource/flint-spark-integration-assembly-0.1.0-SNAPSHOT.jar";
+  public static final String FLINT_DEFAULT_HOST = "localhost";
+  public static final String FLINT_DEFAULT_PORT = "9200";
+  public static final String FLINT_DEFAULT_SCHEME = "http";
+  public static final String FLINT_DEFAULT_AUTH = "-1";
+  public static final String FLINT_DEFAULT_REGION = "us-west-2";
+}