Merge pull request #450 from qiaozhanwei/dev-1.1.0

hdfs add kerberos authentication

escheduler-common/src/main/java/cn/escheduler/common/Constants.java

@@ -241,7 +241,11 @@ public final class Constants {
      */
     public static final String SCHEDULER_TASKS_QUEUE = "tasks_queue";
 
+    /**
+     * escheduler need kill tasks queue
+     */
     public static final String SCHEDULER_TASKS_KILL = "tasks_kill";
+
     public static final String ZOOKEEPER_SCHEDULER_ROOT = "zookeeper.escheduler.root";
 
     public static final String SCHEDULER_QUEUE_IMPL = "escheduler.queue.impl";
@@ -336,11 +340,6 @@ public final class Constants {
      */
     public static final int MAX_TASK_TIMEOUT = 24 * 3600;
 
-    /**
-     * max task timeout
-     */
-    public static final int MAX_PROCESS_TIMEOUT = Integer.MAX_VALUE;
-
 
     /**
      * heartbeat threads number
@@ -830,4 +829,35 @@ public final class Constants {
      *  preview schedule execute count
      */
     public static final int PREVIEW_SCHEDULE_EXECUTE_COUNT = 5;
+
+    /**
+     * java.security.krb5.conf
+     */
+    public static final String JAVA_SECURITY_KRB5_CONF = "java.security.krb5.conf";
+
+    /**
+     * java.security.krb5.conf.path
+     */
+    public static final String JAVA_SECURITY_KRB5_CONF_PATH = "java.security.krb5.conf.path";
+
+    /**
+     * hadoop.security.authentication
+     */
+    public static final String HADOOP_SECURITY_AUTHENTICATION = "hadoop.security.authentication";
+
+    /**
+     * hadoop.security.authentication
+     */
+    public static final String HADOOP_SECURITY_AUTHENTICATION_STARTUP_STATE = "hadoop.security.authentication.startup.state";
+
+
+    /**
+     * loginUserFromKeytab user
+     */
+    public static final String LOGIN_USER_KEY_TAB_USERNAME = "login.user.keytab.username";
+
+    /**
+     * loginUserFromKeytab path
+     */
+    public static final String LOGIN_USER_KEY_TAB_PATH = "login.user.keytab.path";
 }

escheduler-common/src/main/java/cn/escheduler/common/utils/HadoopUtils.java

@@ -29,6 +29,7 @@ import org.apache.hadoop.fs.FileSystem;
 import org.apache.hadoop.fs.FileUtil;
 import org.apache.hadoop.fs.Path;
 import org.apache.hadoop.fs.FileStatus;
+import org.apache.hadoop.security.UserGroupInformation;
 import org.apache.hadoop.yarn.client.cli.RMAdminCLI;
 import org.slf4j.Logger;
 import org.slf4j.LoggerFactory;
@@ -40,9 +41,7 @@ import java.util.stream.Collectors;
 import java.util.stream.Stream;
 
 import static cn.escheduler.common.Constants.*;
-import static cn.escheduler.common.utils.PropertyUtils.getInt;
+import static cn.escheduler.common.utils.PropertyUtils.*;
-import static cn.escheduler.common.utils.PropertyUtils.getString;
-import static cn.escheduler.common.utils.PropertyUtils.getPrefixedProperties;
 
 /**
  * hadoop utils
@@ -73,6 +72,16 @@ public class HadoopUtils implements Closeable {
                 if (configuration == null) {
                     try {
                         configuration = new Configuration();
+
+                        if (getBoolean(Constants.HADOOP_SECURITY_AUTHENTICATION_STARTUP_STATE)){
+                            System.setProperty(Constants.JAVA_SECURITY_KRB5_CONF,
+                                    getString(Constants.JAVA_SECURITY_KRB5_CONF_PATH));
+                            configuration.set(Constants.HADOOP_SECURITY_AUTHENTICATION,"kerberos");
+                            UserGroupInformation.setConfiguration(configuration);
+                            UserGroupInformation.loginUserFromKeytab(getString(Constants.LOGIN_USER_KEY_TAB_USERNAME),
+                                    getString(Constants.LOGIN_USER_KEY_TAB_PATH));
+                        }
+
                         String defaultFS = configuration.get(FS_DEFAULTFS);
                         //first get key from core-site.xml hdfs-site.xml ,if null ,then try to get from properties file
                         // the default is the local file system
@@ -155,7 +164,7 @@ public class HadoopUtils implements Closeable {
      */
     public List<String> catFile(String hdfsFilePath, int skipLineNums, int limit) throws IOException {
 
-        if(StringUtils.isBlank(hdfsFilePath)){
+        if (StringUtils.isBlank(hdfsFilePath)){
             logger.error("hdfs file path:{} is blank",hdfsFilePath);
             return null;
         }

escheduler-common/src/main/resources/common/common.properties

@@ -14,7 +14,19 @@ process.exec.basepath=/tmp/escheduler/exec
 data.store2hdfs.basepath=/escheduler
 
 # whether hdfs starts
-hdfs.startup.state=true
+hdfs.startup.state=false
+
+# whether kerberos starts
+hadoop.security.authentication.startup.state=false
+
+# java.security.krb5.conf path
+java.security.krb5.conf.path=/opt/krb5.conf
+
+# loginUserFromKeytab user
+login.user.keytab.username="hdfs-mycluster@ESZ.COM"
+
+# loginUserFromKeytab path
+login.user.keytab.path="/opt/hdfs.headless.keytab"
 
 # system env path. self configuration, please make sure the directory and file exists and have read write execute permissions
 escheduler.env.path=/opt/.escheduler_env.sh
@@ -23,5 +35,5 @@ escheduler.env.path=/opt/.escheduler_env.sh
 resource.view.suffixs=txt,log,sh,conf,cfg,py,java,sql,hql,xml
 
 # is development state? default "false"
-development.state=false
+development.state=true
 

install.sh

@@ -125,7 +125,7 @@ yarnHaIps="192.168.xx.xx,192.168.xx.xx"
 # 如果是单 resourcemanager,只需要配置一个主机名称,如果是resourcemanager HA,则默认配置就好
 singleYarnIp="ark1"
 
-# hdfs根路径,根路径的owner必须是部署用户
+# hdfs根路径，根路径的owner必须是部署用户。1.1.0之前版本不会自动创建hdfs根目录，需要自行创建
 hdfsPath="/escheduler"
 
 # common 配置
@@ -147,6 +147,19 @@ resSuffixs="txt,log,sh,conf,cfg,py,java,sql,hql,xml"
 # 开发状态,如果是true,对于SHELL脚本可以在execPath目录下查看封装后的SHELL脚本,如果是false则执行完成直接删除
 devState="true"
 
+# kerberos 配置
+# kerberos 是否启动
+kerberosStartUp="false"
+
+# kdc krb5 配置文件路径
+krb5ConfPath="$installPath/conf/krb5.conf"
+
+# keytab 用户名
+keytabUserName="hdfs-mycluster@ESZ.COM"
+
+# 用户 keytab路径
+keytabPath="$installPath/conf/hdfs.headless.keytab"
+
 # zk 配置
 # zk根目录
 zkRoot="/escheduler"