You are viewing a plain text version of this content. The canonical link for it is here.
Posted to dev@zeppelin.apache.org by GitBox <gi...@apache.org> on 2021/04/19 05:22:12 UTC
[GitHub] [zeppelin] zjffdu opened a new pull request #4098: [ZEPPELIN-5320]. Support yarn application mode for flink interpreter
zjffdu opened a new pull request #4098:
URL: https://github.com/apache/zeppelin/pull/4098
### What is this PR for?
This PR is to support yarn application mode of flink interpreter. In yarn application mode, flink interpreter is running in JM. Integration test is added.
### What type of PR is it?
[Feature | Documentation ]
### Todos
* [ ] - Task
### What is the Jira issue?
* https://issues.apache.org/jira/browse/ZEPPELIN-5320
### How should this be tested?
* CI pass
### Screenshots (if appropriate)
### Questions:
* Does the licenses files need update? No
* Is there breaking changes for older versions? No
* Does this needs documentation? No
--
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:
users@infra.apache.org
[GitHub] [zeppelin] zjffdu commented on pull request #4098: [ZEPPELIN-5320]. Support yarn application mode for flink interpreter
Posted by GitBox <gi...@apache.org>.
zjffdu commented on pull request #4098:
URL: https://github.com/apache/zeppelin/pull/4098#issuecomment-830637778
Will merge if no more comment
--
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:
users@infra.apache.org
[GitHub] [zeppelin] zjffdu commented on a change in pull request #4098: [ZEPPELIN-5320]. Support yarn application mode for flink interpreter
Posted by GitBox <gi...@apache.org>.
zjffdu commented on a change in pull request #4098:
URL: https://github.com/apache/zeppelin/pull/4098#discussion_r615960950
##########
File path: flink/interpreter/src/main/java/org/apache/zeppelin/flink/YarnApplicationExecutionEnvironment.java
##########
@@ -0,0 +1,82 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.zeppelin.flink;
+
+import org.apache.flink.api.common.JobExecutionResult;
+import org.apache.flink.api.java.ExecutionEnvironment;
+import org.apache.flink.configuration.ConfigUtils;
+import org.apache.flink.configuration.Configuration;
+import org.apache.flink.configuration.DeploymentOptions;
+import org.apache.flink.configuration.PipelineOptions;
+import org.apache.flink.core.execution.JobClient;
+import org.apache.flink.core.execution.PipelineExecutorServiceLoader;
+
+import java.net.MalformedURLException;
+import java.net.URL;
+import java.util.ArrayList;
+import java.util.List;
+
+import static org.apache.flink.util.Preconditions.checkState;
+
+
+/**
+ * ExecutionEnvironment used for yarn application mode.
+ * Need to add jars of scala shell before submitting jobs.
+ */
+public class YarnApplicationExecutionEnvironment extends ExecutionEnvironment {
+
+ private FlinkILoop flinkILoop;
+
+ public YarnApplicationExecutionEnvironment(PipelineExecutorServiceLoader executorServiceLoader,
+ Configuration configuration,
+ ClassLoader userClassloader,
+ FlinkILoop flinkILoop) {
+ super(executorServiceLoader,configuration,userClassloader);
+ this.flinkILoop = flinkILoop;
+ }
+
+ @Override
+ public JobClient executeAsync(String jobName) throws Exception {
+ updateDependencies();
+ return super.executeAsync(jobName);
+ }
+
+ @Override
+ public JobExecutionResult execute() throws Exception {
+ updateDependencies();
+ return super.execute();
+ }
+
+ private void updateDependencies() throws Exception {
+ final Configuration configuration = getConfiguration();
+ checkState(
+ configuration.getBoolean(DeploymentOptions.ATTACHED),
+ "Only ATTACHED mode is supported by the scala shell.");
+
+ final List<URL> updatedJarFiles = getUpdatedJarFiles();
+ ConfigUtils.encodeCollectionToConfig(
+ configuration, PipelineOptions.JARS, updatedJarFiles, URL::toString);
+ }
+
+ private List<URL> getUpdatedJarFiles() throws MalformedURLException {
+ final URL jarUrl = flinkILoop.writeFilesToDisk().getAbsoluteFile().toURI().toURL();
+ final List<URL> allJarFiles = new ArrayList<>();
+ allJarFiles.add(jarUrl);
+ return allJarFiles;
+ }
Review comment:
I think it is not side-effects, it just generate jar files, but didn't change the internal state of `YarnApplicationExecutionEnvironment`
--
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:
users@infra.apache.org
[GitHub] [zeppelin] cuspymd commented on a change in pull request #4098: [ZEPPELIN-5320]. Support yarn application mode for flink interpreter
Posted by GitBox <gi...@apache.org>.
cuspymd commented on a change in pull request #4098:
URL: https://github.com/apache/zeppelin/pull/4098#discussion_r616012945
##########
File path: zeppelin-plugins/launcher/flink/src/main/java/org/apache/zeppelin/interpreter/launcher/FlinkInterpreterLauncher.java
##########
@@ -55,6 +64,65 @@ public FlinkInterpreterLauncher(ZeppelinConfiguration zConf, RecoveryStorage rec
}
envs.put("FLINK_LIB_DIR", flinkHome + "/lib");
envs.put("FLINK_PLUGINS_DIR", flinkHome + "/plugins");
+
+ // yarn application mode specific logic
+ if (context.getProperties().getProperty("flink.execution.mode")
+ .equalsIgnoreCase("yarn_application")) {
+ updateEnvsForYarnApplicationMode(envs, context);
+ }
+
return envs;
}
+
+ private void updateEnvsForYarnApplicationMode(Map<String, String> envs,
+ InterpreterLaunchContext context) {
+ envs.put("ZEPPELIN_FLINK_YARN_APPLICATION", "true");
+
+ StringBuilder flinkYarnApplicationConfBuilder = new StringBuilder();
+
+ // Extract yarn.ship-files, add hive-site.xml automatically if hive is enabled
+ // and HIVE_CONF_DIR is specified
+ String hiveConfDirProperty = context.getProperties().getProperty("HIVE_CONF_DIR");
+ List<String> yarnShipFiles = new ArrayList<>();
+ if (StringUtils.isNotBlank(hiveConfDirProperty) &&
+ Boolean.parseBoolean(context.getProperties()
+ .getProperty("zeppelin.flink.enableHive", "false"))) {
+ File hiveSiteFile = new File(hiveConfDirProperty, "hive-site.xml");
+ if (hiveSiteFile.isFile() && hiveSiteFile.exists()) {
+ yarnShipFiles.add(hiveSiteFile.getAbsolutePath());
+ } else {
+ LOGGER.warn("Hive site file: {} doesn't exist or is not a directory", hiveSiteFile);
+ }
+ }
+ if (context.getProperties().containsKey("yarn.ship-files")) {
+ yarnShipFiles.add(context.getProperties().getProperty("yarn.ship-files"));
+ }
Review comment:
Separation with `getYarnShipFiles()` function seems to improve readability.
##########
File path: zeppelin-plugins/launcher/flink/src/main/java/org/apache/zeppelin/interpreter/launcher/FlinkInterpreterLauncher.java
##########
@@ -55,6 +64,65 @@ public FlinkInterpreterLauncher(ZeppelinConfiguration zConf, RecoveryStorage rec
}
envs.put("FLINK_LIB_DIR", flinkHome + "/lib");
envs.put("FLINK_PLUGINS_DIR", flinkHome + "/plugins");
+
+ // yarn application mode specific logic
+ if (context.getProperties().getProperty("flink.execution.mode")
+ .equalsIgnoreCase("yarn_application")) {
+ updateEnvsForYarnApplicationMode(envs, context);
+ }
+
return envs;
}
+
+ private void updateEnvsForYarnApplicationMode(Map<String, String> envs,
+ InterpreterLaunchContext context) {
+ envs.put("ZEPPELIN_FLINK_YARN_APPLICATION", "true");
+
+ StringBuilder flinkYarnApplicationConfBuilder = new StringBuilder();
+
+ // Extract yarn.ship-files, add hive-site.xml automatically if hive is enabled
+ // and HIVE_CONF_DIR is specified
+ String hiveConfDirProperty = context.getProperties().getProperty("HIVE_CONF_DIR");
+ List<String> yarnShipFiles = new ArrayList<>();
+ if (StringUtils.isNotBlank(hiveConfDirProperty) &&
+ Boolean.parseBoolean(context.getProperties()
+ .getProperty("zeppelin.flink.enableHive", "false"))) {
+ File hiveSiteFile = new File(hiveConfDirProperty, "hive-site.xml");
+ if (hiveSiteFile.isFile() && hiveSiteFile.exists()) {
+ yarnShipFiles.add(hiveSiteFile.getAbsolutePath());
+ } else {
+ LOGGER.warn("Hive site file: {} doesn't exist or is not a directory", hiveSiteFile);
+ }
+ }
+ if (context.getProperties().containsKey("yarn.ship-files")) {
+ yarnShipFiles.add(context.getProperties().getProperty("yarn.ship-files"));
+ }
+ if (!yarnShipFiles.isEmpty()) {
+ flinkYarnApplicationConfBuilder.append(
+ " -D yarn.ship-files=" + yarnShipFiles.stream().collect(Collectors.joining(",")));
+ }
+
+ // specify yarn.application.name
+ String yarnAppName = context.getProperties().getProperty("flink.yarn.appName");
+ if (StringUtils.isNotBlank(yarnAppName)) {
+ // flink run command can not contains whitespace, so replace it with _
+ flinkYarnApplicationConfBuilder.append(
+ " -D yarn.application.name=" + yarnAppName.replaceAll(" ", "_") + "");
+ }
+
+ // add other yarn and python configuration.
+ for (Map.Entry<Object, Object> entry : context.getProperties().entrySet()) {
+ if (!entry.getKey().toString().equalsIgnoreCase("yarn.ship-files") &&
+ !entry.getKey().toString().equalsIgnoreCase("flink.yarn.appName")) {
+ if (CharMatcher.whitespace().matchesAnyOf(entry.getValue().toString())) {
+ LOGGER.warn("flink configuration key {} is skipped because it contains white space",
+ entry.getValue().toString());
+ } else {
+ flinkYarnApplicationConfBuilder.append(
+ " -D " + entry.getKey().toString() + "=" + entry.getValue().toString() + "");
+ }
+ }
+ }
Review comment:
`entry.getKey().toString()` and `entry.getValue().toString()` are repeated several times, so it seems to be good to use by explicitly declaring them as variables.
--
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:
users@infra.apache.org
[GitHub] [zeppelin] zjffdu commented on a change in pull request #4098: [ZEPPELIN-5320]. Support yarn application mode for flink interpreter
Posted by GitBox <gi...@apache.org>.
zjffdu commented on a change in pull request #4098:
URL: https://github.com/apache/zeppelin/pull/4098#discussion_r615960950
##########
File path: flink/interpreter/src/main/java/org/apache/zeppelin/flink/YarnApplicationExecutionEnvironment.java
##########
@@ -0,0 +1,82 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.zeppelin.flink;
+
+import org.apache.flink.api.common.JobExecutionResult;
+import org.apache.flink.api.java.ExecutionEnvironment;
+import org.apache.flink.configuration.ConfigUtils;
+import org.apache.flink.configuration.Configuration;
+import org.apache.flink.configuration.DeploymentOptions;
+import org.apache.flink.configuration.PipelineOptions;
+import org.apache.flink.core.execution.JobClient;
+import org.apache.flink.core.execution.PipelineExecutorServiceLoader;
+
+import java.net.MalformedURLException;
+import java.net.URL;
+import java.util.ArrayList;
+import java.util.List;
+
+import static org.apache.flink.util.Preconditions.checkState;
+
+
+/**
+ * ExecutionEnvironment used for yarn application mode.
+ * Need to add jars of scala shell before submitting jobs.
+ */
+public class YarnApplicationExecutionEnvironment extends ExecutionEnvironment {
+
+ private FlinkILoop flinkILoop;
+
+ public YarnApplicationExecutionEnvironment(PipelineExecutorServiceLoader executorServiceLoader,
+ Configuration configuration,
+ ClassLoader userClassloader,
+ FlinkILoop flinkILoop) {
+ super(executorServiceLoader,configuration,userClassloader);
+ this.flinkILoop = flinkILoop;
+ }
+
+ @Override
+ public JobClient executeAsync(String jobName) throws Exception {
+ updateDependencies();
+ return super.executeAsync(jobName);
+ }
+
+ @Override
+ public JobExecutionResult execute() throws Exception {
+ updateDependencies();
+ return super.execute();
+ }
+
+ private void updateDependencies() throws Exception {
+ final Configuration configuration = getConfiguration();
+ checkState(
+ configuration.getBoolean(DeploymentOptions.ATTACHED),
+ "Only ATTACHED mode is supported by the scala shell.");
+
+ final List<URL> updatedJarFiles = getUpdatedJarFiles();
+ ConfigUtils.encodeCollectionToConfig(
+ configuration, PipelineOptions.JARS, updatedJarFiles, URL::toString);
+ }
+
+ private List<URL> getUpdatedJarFiles() throws MalformedURLException {
+ final URL jarUrl = flinkILoop.writeFilesToDisk().getAbsoluteFile().toURI().toURL();
+ final List<URL> allJarFiles = new ArrayList<>();
+ allJarFiles.add(jarUrl);
+ return allJarFiles;
+ }
Review comment:
I think it is not side-effects, it just generate jar files, but doesn't change the internal state of `YarnApplicationExecutionEnvironment`
##########
File path: zeppelin-plugins/launcher/flink/src/main/java/org/apache/zeppelin/interpreter/launcher/FlinkInterpreterLauncher.java
##########
@@ -55,6 +64,60 @@ public FlinkInterpreterLauncher(ZeppelinConfiguration zConf, RecoveryStorage rec
}
envs.put("FLINK_LIB_DIR", flinkHome + "/lib");
envs.put("FLINK_PLUGINS_DIR", flinkHome + "/plugins");
+
+ // yarn application mode specific logic
+ if (context.getProperties().getProperty("flink.execution.mode")
+ .equalsIgnoreCase("yarn_application")) {
+ envs.put("ZEPPELIN_FLINK_YARN_APPLICATION", "true");
+
+ StringBuilder flinkYarnApplicationConfBuilder = new StringBuilder();
+
+ // Extract yarn.ship-files, add hive-site.xml automatically if hive is enabled
+ // and HIVE_CONF_DIR is specified
+ String hiveConfDirProperty = context.getProperties().getProperty("HIVE_CONF_DIR");
+ List<String> yarnShipFiles = new ArrayList<>();
+ if (StringUtils.isNotBlank(hiveConfDirProperty) &&
+ Boolean.parseBoolean(context.getProperties()
+ .getProperty("zeppelin.flink.enableHive", "false"))) {
+ File hiveSiteFile = new File(hiveConfDirProperty, "hive-site.xml");
+ if (hiveSiteFile.isFile() && hiveSiteFile.exists()) {
+ yarnShipFiles.add(hiveSiteFile.getAbsolutePath());
+ } else {
+ LOGGER.warn("Hive site file: {} doesn't exist or is not a directory", hiveSiteFile);
+ }
+ }
+ if (context.getProperties().containsKey("yarn.ship-files")) {
+ yarnShipFiles.add(context.getProperties().getProperty("yarn.ship-files"));
+ }
+ if (!yarnShipFiles.isEmpty()) {
+ flinkYarnApplicationConfBuilder.append(
+ " -D yarn.ship-files=" + yarnShipFiles.stream().collect(Collectors.joining(",")));
+ }
+
+ // specify yarn.application.name
+ String yarnAppName = context.getProperties().getProperty("flink.yarn.appName");
+ if (StringUtils.isNotBlank(yarnAppName)) {
+ // flink run command can not contains whitespace, so replace it with _
+ flinkYarnApplicationConfBuilder.append(
+ " -D yarn.application.name=" + yarnAppName.replaceAll(" ", "_") + "");
+ }
+
+ // add other yarn and python configuration.
+ for (Map.Entry<Object, Object> entry : context.getProperties().entrySet()) {
+ if (!entry.getKey().toString().equalsIgnoreCase("yarn.ship-files") &&
+ !entry.getKey().toString().equalsIgnoreCase("flink.yarn.appName")) {
+ if (CharMatcher.whitespace().matchesAnyOf(entry.getValue().toString())) {
+ LOGGER.warn("flink configuration key {} is skipped because it contains white space",
+ entry.getValue().toString());
+ } else {
+ flinkYarnApplicationConfBuilder.append(
+ " -D " + entry.getKey().toString() + "=" + entry.getValue().toString() + "");
+ }
+ }
+ }
+ envs.put("ZEPPELIN_FLINK_YANR_APPLICATION_CONF", flinkYarnApplicationConfBuilder.toString());
+ }
Review comment:
Good point, fixed
##########
File path: zeppelin-zengine/src/main/java/org/apache/zeppelin/interpreter/remote/ExecRemoteInterpreterProcess.java
##########
@@ -215,6 +215,14 @@ public void onProcessComplete(int exitValue) {
notifyAll();
}
}
+ if (getEnv().getOrDefault("ZEPPELIN_FLINK_YARN_APPLICATION", "false").equalsIgnoreCase("true")
Review comment:
Fixed
--
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:
users@infra.apache.org
[GitHub] [zeppelin] zjffdu commented on a change in pull request #4098: [ZEPPELIN-5320]. Support yarn application mode for flink interpreter
Posted by GitBox <gi...@apache.org>.
zjffdu commented on a change in pull request #4098:
URL: https://github.com/apache/zeppelin/pull/4098#discussion_r616455596
##########
File path: zeppelin-plugins/launcher/flink/src/main/java/org/apache/zeppelin/interpreter/launcher/FlinkInterpreterLauncher.java
##########
@@ -55,6 +64,65 @@ public FlinkInterpreterLauncher(ZeppelinConfiguration zConf, RecoveryStorage rec
}
envs.put("FLINK_LIB_DIR", flinkHome + "/lib");
envs.put("FLINK_PLUGINS_DIR", flinkHome + "/plugins");
+
+ // yarn application mode specific logic
+ if (context.getProperties().getProperty("flink.execution.mode")
+ .equalsIgnoreCase("yarn_application")) {
+ updateEnvsForYarnApplicationMode(envs, context);
+ }
+
return envs;
}
+
+ private void updateEnvsForYarnApplicationMode(Map<String, String> envs,
+ InterpreterLaunchContext context) {
+ envs.put("ZEPPELIN_FLINK_YARN_APPLICATION", "true");
+
+ StringBuilder flinkYarnApplicationConfBuilder = new StringBuilder();
+
+ // Extract yarn.ship-files, add hive-site.xml automatically if hive is enabled
+ // and HIVE_CONF_DIR is specified
+ String hiveConfDirProperty = context.getProperties().getProperty("HIVE_CONF_DIR");
+ List<String> yarnShipFiles = new ArrayList<>();
+ if (StringUtils.isNotBlank(hiveConfDirProperty) &&
+ Boolean.parseBoolean(context.getProperties()
+ .getProperty("zeppelin.flink.enableHive", "false"))) {
+ File hiveSiteFile = new File(hiveConfDirProperty, "hive-site.xml");
+ if (hiveSiteFile.isFile() && hiveSiteFile.exists()) {
+ yarnShipFiles.add(hiveSiteFile.getAbsolutePath());
+ } else {
+ LOGGER.warn("Hive site file: {} doesn't exist or is not a directory", hiveSiteFile);
+ }
+ }
+ if (context.getProperties().containsKey("yarn.ship-files")) {
+ yarnShipFiles.add(context.getProperties().getProperty("yarn.ship-files"));
+ }
+ if (!yarnShipFiles.isEmpty()) {
+ flinkYarnApplicationConfBuilder.append(
+ " -D yarn.ship-files=" + yarnShipFiles.stream().collect(Collectors.joining(",")));
+ }
+
+ // specify yarn.application.name
+ String yarnAppName = context.getProperties().getProperty("flink.yarn.appName");
+ if (StringUtils.isNotBlank(yarnAppName)) {
+ // flink run command can not contains whitespace, so replace it with _
+ flinkYarnApplicationConfBuilder.append(
+ " -D yarn.application.name=" + yarnAppName.replaceAll(" ", "_") + "");
+ }
+
+ // add other yarn and python configuration.
+ for (Map.Entry<Object, Object> entry : context.getProperties().entrySet()) {
+ if (!entry.getKey().toString().equalsIgnoreCase("yarn.ship-files") &&
+ !entry.getKey().toString().equalsIgnoreCase("flink.yarn.appName")) {
+ if (CharMatcher.whitespace().matchesAnyOf(entry.getValue().toString())) {
+ LOGGER.warn("flink configuration key {} is skipped because it contains white space",
+ entry.getValue().toString());
+ } else {
+ flinkYarnApplicationConfBuilder.append(
+ " -D " + entry.getKey().toString() + "=" + entry.getValue().toString() + "");
+ }
+ }
+ }
Review comment:
Thanks for the careful review @cuspymd
Comments are addressed
--
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:
users@infra.apache.org
[GitHub] [zeppelin] zjffdu commented on a change in pull request #4098: [ZEPPELIN-5320]. Support yarn application mode for flink interpreter
Posted by GitBox <gi...@apache.org>.
zjffdu commented on a change in pull request #4098:
URL: https://github.com/apache/zeppelin/pull/4098#discussion_r618078514
##########
File path: docs/interpreter/flink.md
##########
@@ -266,6 +266,7 @@ Flink in Zeppelin supports 3 execution modes (`flink.execution.mode`):
* Local
* Remote
* Yarn
+* Yarn Application
Review comment:
Fixed
--
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:
users@infra.apache.org
[GitHub] [zeppelin] asfgit closed pull request #4098: [ZEPPELIN-5320]. Support yarn application mode for flink interpreter
Posted by GitBox <gi...@apache.org>.
asfgit closed pull request #4098:
URL: https://github.com/apache/zeppelin/pull/4098
--
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:
users@infra.apache.org
[GitHub] [zeppelin] cuspymd commented on a change in pull request #4098: [ZEPPELIN-5320]. Support yarn application mode for flink interpreter
Posted by GitBox <gi...@apache.org>.
cuspymd commented on a change in pull request #4098:
URL: https://github.com/apache/zeppelin/pull/4098#discussion_r615706250
##########
File path: zeppelin-zengine/src/main/java/org/apache/zeppelin/interpreter/remote/ExecRemoteInterpreterProcess.java
##########
@@ -215,6 +215,14 @@ public void onProcessComplete(int exitValue) {
notifyAll();
}
}
+ if (getEnv().getOrDefault("ZEPPELIN_FLINK_YARN_APPLICATION", "false").equalsIgnoreCase("true")
Review comment:
Isn't the `else if` clearer?
##########
File path: flink/interpreter/src/main/java/org/apache/zeppelin/flink/YarnApplicationExecutionEnvironment.java
##########
@@ -0,0 +1,82 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one or more
+ * contributor license agreements. See the NOTICE file distributed with
+ * this work for additional information regarding copyright ownership.
+ * The ASF licenses this file to You under the Apache License, Version 2.0
+ * (the "License"); you may not use this file except in compliance with
+ * the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing, software
+ * distributed under the License is distributed on an "AS IS" BASIS,
+ * WITHOUT WARRANTIES OR CONDITIONS OF ANY KIND, either express or implied.
+ * See the License for the specific language governing permissions and
+ * limitations under the License.
+ */
+
+package org.apache.zeppelin.flink;
+
+import org.apache.flink.api.common.JobExecutionResult;
+import org.apache.flink.api.java.ExecutionEnvironment;
+import org.apache.flink.configuration.ConfigUtils;
+import org.apache.flink.configuration.Configuration;
+import org.apache.flink.configuration.DeploymentOptions;
+import org.apache.flink.configuration.PipelineOptions;
+import org.apache.flink.core.execution.JobClient;
+import org.apache.flink.core.execution.PipelineExecutorServiceLoader;
+
+import java.net.MalformedURLException;
+import java.net.URL;
+import java.util.ArrayList;
+import java.util.List;
+
+import static org.apache.flink.util.Preconditions.checkState;
+
+
+/**
+ * ExecutionEnvironment used for yarn application mode.
+ * Need to add jars of scala shell before submitting jobs.
+ */
+public class YarnApplicationExecutionEnvironment extends ExecutionEnvironment {
+
+ private FlinkILoop flinkILoop;
+
+ public YarnApplicationExecutionEnvironment(PipelineExecutorServiceLoader executorServiceLoader,
+ Configuration configuration,
+ ClassLoader userClassloader,
+ FlinkILoop flinkILoop) {
+ super(executorServiceLoader,configuration,userClassloader);
+ this.flinkILoop = flinkILoop;
+ }
+
+ @Override
+ public JobClient executeAsync(String jobName) throws Exception {
+ updateDependencies();
+ return super.executeAsync(jobName);
+ }
+
+ @Override
+ public JobExecutionResult execute() throws Exception {
+ updateDependencies();
+ return super.execute();
+ }
+
+ private void updateDependencies() throws Exception {
+ final Configuration configuration = getConfiguration();
+ checkState(
+ configuration.getBoolean(DeploymentOptions.ATTACHED),
+ "Only ATTACHED mode is supported by the scala shell.");
+
+ final List<URL> updatedJarFiles = getUpdatedJarFiles();
+ ConfigUtils.encodeCollectionToConfig(
+ configuration, PipelineOptions.JARS, updatedJarFiles, URL::toString);
+ }
+
+ private List<URL> getUpdatedJarFiles() throws MalformedURLException {
+ final URL jarUrl = flinkILoop.writeFilesToDisk().getAbsoluteFile().toURI().toURL();
+ final List<URL> allJarFiles = new ArrayList<>();
+ allJarFiles.add(jarUrl);
+ return allJarFiles;
+ }
Review comment:
It's not a simple getter, but a function that generates side-effects. Isn't `updateJarFiles()` a more accurate name?
##########
File path: flink/interpreter/src/main/java/org/apache/zeppelin/flink/HadoopUtils.java
##########
@@ -55,6 +65,18 @@ public static String getYarnAppTrackingUrl(ClusterClient clusterClient) throws I
return yarnClient.getApplicationReport(yarnAppId).getTrackingUrl();
}
+ public static int getFlinkRestPort(String yarnAppId) throws IOException, YarnException {
+ YarnClient yarnClient = YarnClient.createYarnClient();
+ YarnConfiguration yarnConf = new YarnConfiguration();
+ // disable timeline service as we only query yarn app here.
+ // Otherwise we may hit this kind of ERROR:
+ // java.lang.ClassNotFoundException: com.sun.jersey.api.client.config.ClientConfig
+ yarnConf.set("yarn.timeline-service.enabled", "false");
+ yarnClient.init(yarnConf);
+ yarnClient.start();
+ return yarnClient.getApplicationReport(ConverterUtils.toApplicationId(yarnAppId)).getRpcPort();
Review comment:
It is almost the same as the function above, but it would be good to separate it with the `getApplicationReport()` function.
##########
File path: zeppelin-plugins/launcher/flink/src/main/java/org/apache/zeppelin/interpreter/launcher/FlinkInterpreterLauncher.java
##########
@@ -55,6 +64,60 @@ public FlinkInterpreterLauncher(ZeppelinConfiguration zConf, RecoveryStorage rec
}
envs.put("FLINK_LIB_DIR", flinkHome + "/lib");
envs.put("FLINK_PLUGINS_DIR", flinkHome + "/plugins");
+
+ // yarn application mode specific logic
+ if (context.getProperties().getProperty("flink.execution.mode")
+ .equalsIgnoreCase("yarn_application")) {
+ envs.put("ZEPPELIN_FLINK_YARN_APPLICATION", "true");
+
+ StringBuilder flinkYarnApplicationConfBuilder = new StringBuilder();
+
+ // Extract yarn.ship-files, add hive-site.xml automatically if hive is enabled
+ // and HIVE_CONF_DIR is specified
+ String hiveConfDirProperty = context.getProperties().getProperty("HIVE_CONF_DIR");
+ List<String> yarnShipFiles = new ArrayList<>();
+ if (StringUtils.isNotBlank(hiveConfDirProperty) &&
+ Boolean.parseBoolean(context.getProperties()
+ .getProperty("zeppelin.flink.enableHive", "false"))) {
+ File hiveSiteFile = new File(hiveConfDirProperty, "hive-site.xml");
+ if (hiveSiteFile.isFile() && hiveSiteFile.exists()) {
+ yarnShipFiles.add(hiveSiteFile.getAbsolutePath());
+ } else {
+ LOGGER.warn("Hive site file: {} doesn't exist or is not a directory", hiveSiteFile);
+ }
+ }
+ if (context.getProperties().containsKey("yarn.ship-files")) {
+ yarnShipFiles.add(context.getProperties().getProperty("yarn.ship-files"));
+ }
+ if (!yarnShipFiles.isEmpty()) {
+ flinkYarnApplicationConfBuilder.append(
+ " -D yarn.ship-files=" + yarnShipFiles.stream().collect(Collectors.joining(",")));
+ }
+
+ // specify yarn.application.name
+ String yarnAppName = context.getProperties().getProperty("flink.yarn.appName");
+ if (StringUtils.isNotBlank(yarnAppName)) {
+ // flink run command can not contains whitespace, so replace it with _
+ flinkYarnApplicationConfBuilder.append(
+ " -D yarn.application.name=" + yarnAppName.replaceAll(" ", "_") + "");
+ }
+
+ // add other yarn and python configuration.
+ for (Map.Entry<Object, Object> entry : context.getProperties().entrySet()) {
+ if (!entry.getKey().toString().equalsIgnoreCase("yarn.ship-files") &&
+ !entry.getKey().toString().equalsIgnoreCase("flink.yarn.appName")) {
+ if (CharMatcher.whitespace().matchesAnyOf(entry.getValue().toString())) {
+ LOGGER.warn("flink configuration key {} is skipped because it contains white space",
+ entry.getValue().toString());
+ } else {
+ flinkYarnApplicationConfBuilder.append(
+ " -D " + entry.getKey().toString() + "=" + entry.getValue().toString() + "");
+ }
+ }
+ }
+ envs.put("ZEPPELIN_FLINK_YANR_APPLICATION_CONF", flinkYarnApplicationConfBuilder.toString());
+ }
Review comment:
Separating the added part into a separate function would be better for maintenance.
##########
File path: flink/flink1.12-shims/src/main/java/org/apache/zeppelin/flink/Flink112Shims.java
##########
@@ -173,13 +173,26 @@ public Object createCatalogManager(Object config) {
@Override
public String getPyFlinkPythonPath(Properties properties) throws IOException {
+ if ("yarn_application".equalsIgnoreCase(properties.getProperty("flink.execution.mode"))) {
+ // for yarn application mode, FLINK_HOME is container working directory
+ String flinkHome = new File(".").getAbsolutePath();
+ List<File> depFiles = null;
+ depFiles = Arrays.asList(new File(flinkHome + "/lib/python").listFiles());
+ StringBuilder builder = new StringBuilder();
+ for (File file : depFiles) {
+ LOGGER.info("Adding extracted file {} to PYTHONPATH", file.getAbsolutePath());
+ builder.append(file.getAbsolutePath() + ":");
+ }
+ return builder.toString();
Review comment:
It is almost the same as the 191~198 line code, so it would be good to separate it with a function.
--
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:
users@infra.apache.org
[GitHub] [zeppelin] zjffdu commented on a change in pull request #4098: [ZEPPELIN-5320]. Support yarn application mode for flink interpreter
Posted by GitBox <gi...@apache.org>.
zjffdu commented on a change in pull request #4098:
URL: https://github.com/apache/zeppelin/pull/4098#discussion_r615960200
##########
File path: flink/flink1.12-shims/src/main/java/org/apache/zeppelin/flink/Flink112Shims.java
##########
@@ -173,13 +173,26 @@ public Object createCatalogManager(Object config) {
@Override
public String getPyFlinkPythonPath(Properties properties) throws IOException {
+ if ("yarn_application".equalsIgnoreCase(properties.getProperty("flink.execution.mode"))) {
+ // for yarn application mode, FLINK_HOME is container working directory
+ String flinkHome = new File(".").getAbsolutePath();
+ List<File> depFiles = null;
+ depFiles = Arrays.asList(new File(flinkHome + "/lib/python").listFiles());
+ StringBuilder builder = new StringBuilder();
+ for (File file : depFiles) {
+ LOGGER.info("Adding extracted file {} to PYTHONPATH", file.getAbsolutePath());
+ builder.append(file.getAbsolutePath() + ":");
+ }
+ return builder.toString();
Review comment:
Good point, Fixed
##########
File path: flink/interpreter/src/main/java/org/apache/zeppelin/flink/HadoopUtils.java
##########
@@ -55,6 +65,18 @@ public static String getYarnAppTrackingUrl(ClusterClient clusterClient) throws I
return yarnClient.getApplicationReport(yarnAppId).getTrackingUrl();
}
+ public static int getFlinkRestPort(String yarnAppId) throws IOException, YarnException {
+ YarnClient yarnClient = YarnClient.createYarnClient();
+ YarnConfiguration yarnConf = new YarnConfiguration();
+ // disable timeline service as we only query yarn app here.
+ // Otherwise we may hit this kind of ERROR:
+ // java.lang.ClassNotFoundException: com.sun.jersey.api.client.config.ClientConfig
+ yarnConf.set("yarn.timeline-service.enabled", "false");
+ yarnClient.init(yarnConf);
+ yarnClient.start();
+ return yarnClient.getApplicationReport(ConverterUtils.toApplicationId(yarnAppId)).getRpcPort();
Review comment:
Good point, fixed
--
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:
users@infra.apache.org
[GitHub] [zeppelin] cuspymd commented on a change in pull request #4098: [ZEPPELIN-5320]. Support yarn application mode for flink interpreter
Posted by GitBox <gi...@apache.org>.
cuspymd commented on a change in pull request #4098:
URL: https://github.com/apache/zeppelin/pull/4098#discussion_r616792488
##########
File path: docs/interpreter/flink.md
##########
@@ -266,6 +266,7 @@ Flink in Zeppelin supports 3 execution modes (`flink.execution.mode`):
* Local
* Remote
* Yarn
+* Yarn Application
Review comment:
Lines 262 and 264 also seem to need correction.
```
## Execution mode (Local/Remote/Yarn)
Flink in Zeppelin supports 3 execution modes (`flink.execution.mode`):
```
--
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:
users@infra.apache.org