worker fault tolerance modify (#2212)

* 1, master persistent task
2. extract  master and worker communication model

* 1, master persistent task
2. extract  master and worker communication model

* 1, master persistent task
2. extract  master and worker communication model

* add license

* modify javadoc error

* TaskExecutionContext create modify

* buildAckCommand taskInstanceId not set modify

* java doc error modify

* add comment

* ExecutorManager interface add generic type

* add TaskInstanceCacheManager receive Worker report result

* TaskInstance setExecutePath

* add TaskInstanceCacheManager to receive Worker Task result report

* TaskInstanceCacheManager add remove method

* add license

* add dispatcht task method

* AbstractCommandExecutor remove db access

* AbstractCommandExecutor remove db access

* AbstractCommandExecutor remove db access

* AbstractCommandExecutor remove db access

* AbstractCommandExecutor remove db access

* AbstractCommandExecutor remove db access

* AbstractCommandExecutor remove db access

* taskInstanceCache is null ,need load from db

* taskInstanceCache is null ,need load from db

* taskInstanceCache is null ,need load from db

* 1，worker TaskPros use TaskExecutionContext replase
2，Master kill Task , KillTaskProcessor modify

* worker remove db

* ShellTask modify

* master persistence processId and appIds

* master persistence processId and appIds

* master add kill task logic

* master add kill task logic

* master add kill task logic

* javadoc error modify

* remove chinese log

* executeDirectly method add Override

* remote module modify

* TaskKillResponseProcessor command type modify

* create buildKillCommand

* host add host:port format

* host add host:port format

* TaskAckProcessor modify

* TaskAckProcessor modify

* task prioriry refator

* remove ITaskQueue

* task prioriry refator

* remove ITaskQueue

* TaskPriority refactor

* remove logs

* WorkerServer refactor

* MasterSchedulerService modify

* WorkerConfig listen port modify

* modify master and worker listen port

* cancelTaskInstance set TaskExecutionContext host,logPath,executePath

* cancelTaskInstance set TaskExecutionContext host,logPath,executePath

* Encapsulate the parameters required by sqltask

* 1，Encapsulate the parameters required by sqltask
2，SQLTask optimization

* AbstractTask modify

* ProcedureTask optimization

* MasterSchedulerService modify

* TaskUpdateQueueConsumer modify

* test

* DataxTask process run debug

* DataxTask process run debug

* add protobuf dependency，MR、Spark task etc need this

* TaskUpdateQueueConsumer modify

* TaskExecutionContextBuilder set TaskInstance workgroup

* WorkerGroupService queryAllGroup modify
query available work group

* 1，get workergroup from zk modify
2，SpringConnectionFactory repeat load modify

* master and worker register ip  use OSUtils.getHost()

* ProcessInstance host set ip:port format

* worker fault tolerance modify

* Constants and .env modify

Co-authored-by: qiaozhanwei <qiaozhanwei@analysys.com.cn>

dolphinscheduler-api/src/main/java/org/apache/dolphinscheduler/api/controller/LoggerController.java

@@ -60,14 +60,14 @@ public class LoggerController extends BaseController {
      */
     @ApiOperation(value = "queryLog", notes= "QUERY_TASK_INSTANCE_LOG_NOTES")
     @ApiImplicitParams({
-            @ApiImplicitParam(name = "taskInstId", value = "TASK_ID", dataType = "Int", example = "100"),
+            @ApiImplicitParam(name = "taskInstanceId", value = "TASK_ID", dataType = "Int", example = "100"),
             @ApiImplicitParam(name = "skipLineNum", value = "SKIP_LINE_NUM", dataType ="Int", example = "100"),
             @ApiImplicitParam(name = "limit", value = "LIMIT", dataType ="Int", example = "100")
     })
     @GetMapping(value = "/detail")
     @ResponseStatus(HttpStatus.OK)
     public Result queryLog(@ApiIgnore @RequestAttribute(value = Constants.SESSION_USER) User loginUser,
-                           @RequestParam(value = "taskInstId") int taskInstanceId,
+                           @RequestParam(value = "taskInstanceId") int taskInstanceId,
                            @RequestParam(value = "skipLineNum") int skipNum,
                            @RequestParam(value = "limit") int limit) {
         try {
@@ -91,12 +91,12 @@ public class LoggerController extends BaseController {
      */
     @ApiOperation(value = "downloadTaskLog", notes= "DOWNLOAD_TASK_INSTANCE_LOG_NOTES")
     @ApiImplicitParams({
-            @ApiImplicitParam(name = "taskInstId", value = "TASK_ID",dataType = "Int", example = "100")
+            @ApiImplicitParam(name = "taskInstanceId", value = "TASK_ID",dataType = "Int", example = "100")
     })
     @GetMapping(value = "/download-log")
     @ResponseBody
     public ResponseEntity downloadTaskLog(@ApiIgnore @RequestAttribute(value = Constants.SESSION_USER) User loginUser,
-                                          @RequestParam(value = "taskInstId") int taskInstanceId) {
+                                          @RequestParam(value = "taskInstanceId") int taskInstanceId) {
         try {
             byte[] logBytes = loggerService.getLogBytes(taskInstanceId);
             return ResponseEntity

dolphinscheduler-common/src/main/java/org/apache/dolphinscheduler/common/Constants.java

@@ -118,20 +118,15 @@ public final class Constants {
      */
     public static final String RES_UPLOAD_STARTUP_TYPE = "res.upload.startup.type";
 
-    /**
-     * zookeeper quorum
-     */
-    public static final String ZOOKEEPER_QUORUM = "zookeeper.quorum";
-
     /**
      * MasterServer directory registered in zookeeper
      */
-    public static final String ZOOKEEPER_DOLPHINSCHEDULER_MASTERS = "/masters";
+    public static final String ZOOKEEPER_DOLPHINSCHEDULER_MASTERS = "/nodes/masters";
 
     /**
      * WorkerServer directory registered in zookeeper
      */
-    public static final String ZOOKEEPER_DOLPHINSCHEDULER_WORKERS = "/workers";
+    public static final String ZOOKEEPER_DOLPHINSCHEDULER_WORKERS = "/nodes/worker";
 
     /**
      * all servers directory registered in zookeeper
@@ -143,10 +138,6 @@ public final class Constants {
      */
     public static final String ZOOKEEPER_DOLPHINSCHEDULER_LOCK_MASTERS = "/lock/masters";
 
-    /**
-     * WorkerServer lock directory registered in zookeeper
-     */
-    public static final String ZOOKEEPER_DOLPHINSCHEDULER_LOCK_WORKERS = "/lock/workers";
 
     /**
      * MasterServer failover directory registered in zookeeper
@@ -163,10 +154,6 @@ public final class Constants {
      */
     public static final String ZOOKEEPER_DOLPHINSCHEDULER_LOCK_FAILOVER_STARTUP_MASTERS = "/lock/failover/startup-masters";
 
-    /**
-     * need send warn times when master server or worker server failover
-     */
-    public static final int DOLPHINSCHEDULER_WARN_TIMES_FAILOVER = 3;
 
     /**
      * comma ,
@@ -203,37 +190,6 @@ public final class Constants {
      */
     public static final String EQUAL_SIGN = "=";
 
-    /**
-     * ZOOKEEPER_SESSION_TIMEOUT
-     */
-    public static final String ZOOKEEPER_SESSION_TIMEOUT = "zookeeper.session.timeout";
-
-    public static final String ZOOKEEPER_CONNECTION_TIMEOUT = "zookeeper.connection.timeout";
-
-    public static final String ZOOKEEPER_RETRY_SLEEP = "zookeeper.retry.sleep";
-    public static final String ZOOKEEPER_RETRY_BASE_SLEEP = "zookeeper.retry.base.sleep";
-    public static final String ZOOKEEPER_RETRY_MAX_SLEEP = "zookeeper.retry.max.sleep";
-
-    public static final String ZOOKEEPER_RETRY_MAXTIME = "zookeeper.retry.maxtime";
-
-
-    public static final String MASTER_HEARTBEAT_INTERVAL = "master.heartbeat.interval";
-
-    public static final String MASTER_EXEC_THREADS = "master.exec.threads";
-
-    public static final String MASTER_EXEC_TASK_THREADS = "master.exec.task.number";
-
-
-    public static final String MASTER_COMMIT_RETRY_TIMES = "master.task.commit.retryTimes";
-
-    public static final String MASTER_COMMIT_RETRY_INTERVAL = "master.task.commit.interval";
-
-
-    public static final String WORKER_EXEC_THREADS = "worker.exec.threads";
-
-    public static final String WORKER_HEARTBEAT_INTERVAL = "worker.heartbeat.interval";
-
-    public static final String WORKER_FETCH_TASK_NUM = "worker.fetch.task.num";
 
     public static final String WORKER_MAX_CPULOAD_AVG = "worker.max.cpuload.avg";
 
@@ -244,17 +200,6 @@ public final class Constants {
     public static final String MASTER_RESERVED_MEMORY = "master.reserved.memory";
 
 
-    /**
-     * dolphinscheduler tasks queue
-     */
-    public static final String DOLPHINSCHEDULER_TASKS_QUEUE = "tasks_queue";
-
-    /**
-     * dolphinscheduler need kill tasks queue
-     */
-    public static final String DOLPHINSCHEDULER_TASKS_KILL = "tasks_kill";
-
-    public static final String ZOOKEEPER_DOLPHINSCHEDULER_ROOT = "zookeeper.dolphinscheduler.root";
 
     public static final String SCHEDULER_QUEUE_IMPL = "dolphinscheduler.queue.impl";
 
@@ -350,26 +295,6 @@ public final class Constants {
     public static final int MAX_TASK_TIMEOUT = 24 * 3600;
 
 
-    /**
-     * heartbeat threads number
-     */
-    public static final int DEFAUL_WORKER_HEARTBEAT_THREAD_NUM = 1;
-
-    /**
-     * heartbeat interval
-     */
-    public static final int DEFAULT_WORKER_HEARTBEAT_INTERVAL = 60;
-
-    /**
-     * worker fetch task number
-     */
-    public static final int DEFAULT_WORKER_FETCH_TASK_NUM = 1;
-
-    /**
-     * worker execute threads number
-     */
-    public static final int DEFAULT_WORKER_EXEC_THREAD_NUM = 10;
-
     /**
      * master cpu load
      */
@@ -391,16 +316,6 @@ public final class Constants {
     public static final double DEFAULT_WORKER_RESERVED_MEMORY = OSUtils.totalMemorySize() / 10;
 
 
-    /**
-     * master execute threads number
-     */
-    public static final int DEFAULT_MASTER_EXEC_THREAD_NUM = 100;
-
-
-    /**
-     * default master concurrent task execute num
-     */
-    public static final int DEFAULT_MASTER_TASK_EXEC_NUM = 20;
 
     /**
      * default log cache rows num,output when reach the number
@@ -408,33 +323,11 @@ public final class Constants {
     public static final int DEFAULT_LOG_ROWS_NUM = 4 * 16;
 
     /**
-     * log flush interval，output when reach the interval
+     * log flush interval?output when reach the interval
      */
     public static final int DEFAULT_LOG_FLUSH_INTERVAL = 1000;
 
 
-    /**
-     * default master heartbeat thread number
-     */
-    public static final int DEFAULT_MASTER_HEARTBEAT_THREAD_NUM = 1;
-
-
-    /**
-     * default master heartbeat interval
-     */
-    public static final int DEFAULT_MASTER_HEARTBEAT_INTERVAL = 60;
-
-    /**
-     * default master commit retry times
-     */
-    public static final int DEFAULT_MASTER_COMMIT_RETRY_TIMES = 5;
-
-
-    /**
-     * default master commit retry interval
-     */
-    public static final int DEFAULT_MASTER_COMMIT_RETRY_INTERVAL = 3000;
-
     /**
      * time unit secong to minutes
      */
@@ -805,7 +698,6 @@ public final class Constants {
     public static final String ALIAS = "alias";
     public static final String CONTENT = "content";
     public static final String DEPENDENT_SPLIT = ":||";
-    public static final String DEPENDENT_ALL = "ALL";
 
 
     /**
@@ -864,7 +756,7 @@ public final class Constants {
      */
     public static final String HIVE_CONF = "hiveconf:";
 
-    //flink 任务
+    //flink ??
     public static final String FLINK_YARN_CLUSTER = "yarn-cluster";
     public static final String FLINK_RUN_MODE = "-m";
     public static final String FLINK_YARN_SLOT = "-ys";
@@ -899,26 +791,20 @@ public final class Constants {
 
     /**
      * data total
-     * 数据总数
      */
     public  static final String COUNT = "count";
 
     /**
      * page size
-     * 每页数据条数
      */
     public  static final String PAGE_SIZE = "pageSize";
 
     /**
      * current page no
-     * 当前页码
      */
     public  static final String PAGE_NUMBER = "pageNo";
 
-    /**
+
-     * result
-     */
-    public static final String RESULT = "result";
 
     /**
      *

dolphinscheduler-server/src/main/java/org/apache/dolphinscheduler/server/builder/TaskExecutionContextBuilder.java

@@ -49,6 +49,7 @@ public class TaskExecutionContextBuilder {
         taskExecutionContext.setExecutePath(taskInstance.getExecutePath());
         taskExecutionContext.setTaskJson(taskInstance.getTaskJson());
         taskExecutionContext.setWorkerGroup(taskInstance.getWorkerGroup());
+        taskExecutionContext.setHost(taskInstance.getHost());
         return this;
     }
 

dolphinscheduler-server/src/main/java/org/apache/dolphinscheduler/server/utils/ProcessUtils.java

@@ -22,6 +22,7 @@ import org.apache.dolphinscheduler.common.utils.LoggerUtils;
 import org.apache.dolphinscheduler.common.utils.OSUtils;
 import org.apache.dolphinscheduler.common.utils.StringUtils;
 import org.apache.commons.io.FileUtils;
+import org.apache.dolphinscheduler.remote.utils.Host;
 import org.apache.dolphinscheduler.server.entity.TaskExecutionContext;
 import org.apache.dolphinscheduler.service.log.LogClientService;
 import org.slf4j.Logger;
@@ -379,7 +380,9 @@ public class ProcessUtils {
       String log = null;
       try {
         logClient = new LogClientService();
-        log = logClient.viewLog(taskExecutionContext.getHost(), Constants.RPC_PORT, taskExecutionContext.getLogPath());
+        log = logClient.viewLog(Host.of(taskExecutionContext.getHost()).getIp(),
+                Constants.RPC_PORT,
+                taskExecutionContext.getLogPath());
       } finally {
         if(logClient != null){
           logClient.close();

dolphinscheduler-server/src/main/java/org/apache/dolphinscheduler/server/zk/ZKMasterClient.java

@@ -71,7 +71,7 @@ public class ZKMasterClient extends AbstractZKClient {
 			// init system znode
 			this.initSystemZNode();
 
-			// check if fault tolerance is required，failure and tolerance
+			// check if fault tolerance is required?failure and tolerance
 			if (getActiveMasterNum() == 1) {
 				failoverWorker(null, true);
 				failoverMaster(null);
@@ -304,7 +304,6 @@ public class ZKMasterClient extends AbstractZKClient {
 			TaskExecutionContext taskExecutionContext = TaskExecutionContextBuilder.get()
 					.buildTaskInstanceRelatedInfo(taskInstance)
 					.buildProcessInstanceRelatedInfo(processInstance)
-					.buildProcessDefinitionRelatedInfo(null)
 					.create();
 			// only kill yarn job if exists , the local thread has exited
 			ProcessUtils.killYarnJob(taskExecutionContext);

dolphinscheduler-service/src/main/java/org/apache/dolphinscheduler/service/zk/AbstractZKClient.java

@@ -38,29 +38,13 @@ public abstract class AbstractZKClient extends ZookeeperCachedOperator {
 
 	private static final Logger logger = LoggerFactory.getLogger(AbstractZKClient.class);
 
+
 	/**
-	 *	check dead server or not , if dead, stop self
+	 *  remove dead server by host
-	 *
+	 * @param host host
-	 * @param zNode   	 node path
+	 * @param serverType serverType
-	 * @param serverType master or worker prefix
+	 * @throws Exception
-	 * @return  true if not exists
-	 * @throws Exception errors
 	 */
-	protected boolean checkIsDeadServer(String zNode, String serverType) throws Exception{
-		//ip_sequenceno
-		String[] zNodesPath = zNode.split("\\/");
-		String ipSeqNo = zNodesPath[zNodesPath.length - 1];
-
-		String type = serverType.equals(MASTER_PREFIX) ? MASTER_PREFIX : WORKER_PREFIX;
-		String deadServerPath = getDeadZNodeParentPath() + SINGLE_SLASH + type + UNDERLINE + ipSeqNo;
-
-		if(!isExisted(zNode) || isExisted(deadServerPath)){
-			return true;
-		}
-		return false;
-	}
-
-
 	public void removeDeadServerByHost(String host, String serverType) throws Exception {
 		List<String> deadServers = super.getChildrenKeys(getDeadZNodeParentPath());
 		for(String serverPath : deadServers){
@@ -72,22 +56,6 @@ public abstract class AbstractZKClient extends ZookeeperCachedOperator {
 		}
 	}
 
-	/**
-	 * create zookeeper path according the zk node type.
-	 * @param zkNodeType zookeeper node type
-	 * @return  register zookeeper path
-	 * @throws Exception
-	 */
-	private String createZNodePath(ZKNodeType zkNodeType, String host) throws Exception {
-		// specify the format of stored data in ZK nodes
-		String heartbeatZKInfo = ResInfo.getHeartBeatInfo(new Date());
-		// create temporary sequence nodes for master znode
-		String registerPath= getZNodeParentPath(zkNodeType) + SINGLE_SLASH + host;
-
-    	super.persistEphemeral(registerPath, heartbeatZKInfo);
-		logger.info("register {} node {} success" , zkNodeType.toString(), registerPath);
-		return registerPath;
-	}
 
 	/**
 	 * opType(add): if find dead server , then add to zk deadServerPath

dolphinscheduler-service/src/main/java/org/apache/dolphinscheduler/service/zk/ZookeeperCachedOperator.java

@@ -39,7 +39,7 @@ public class ZookeeperCachedOperator extends ZookeeperOperator {
      */
     @Override
     protected void registerListener() {
-        treeCache = new TreeCache(zkClient, getZookeeperConfig().getDsRoot());
+        treeCache = new TreeCache(zkClient, getZookeeperConfig().getDsRoot() + "/nodes");
         logger.info("add listener to zk path: {}", getZookeeperConfig().getDsRoot());
         try {
             treeCache.start();