[GitHub] [incubator-dolphinscheduler] yangyichao-mango commented on a change in pull request #2934: [Feature-2815][server] One worker can belong to different workergroups
yangyichao-mango commented on a change in pull request #2934:
URL:
https://github.com/apache/incubator-dolphinscheduler/pull/2934#discussion_r447543546
##
File path:
dolphinscheduler-server/src/main/java/org/apache/dolphinscheduler/server/worker/registry/WorkerRegistry.java
##
@@ -86,34 +90,39 @@ public void init(){
*/
public void registry() {
String address = OSUtils.getHost();
-String localNodePath = getWorkerPath();
-
zookeeperRegistryCenter.getZookeeperCachedOperator().persistEphemeral(localNodePath,
"");
-
zookeeperRegistryCenter.getZookeeperCachedOperator().getZkClient().getConnectionStateListenable().addListener(new
ConnectionStateListener() {
-@Override
-public void stateChanged(CuratorFramework client, ConnectionState
newState) {
-if(newState == ConnectionState.LOST){
-logger.error("worker : {} connection lost from zookeeper",
address);
-} else if(newState == ConnectionState.RECONNECTED){
-logger.info("worker : {} reconnected to zookeeper",
address);
-
zookeeperRegistryCenter.getZookeeperCachedOperator().persistEphemeral(localNodePath,
"");
-} else if(newState == ConnectionState.SUSPENDED){
-logger.warn("worker : {} connection SUSPENDED ", address);
-}
-}
-});
+Set workerZkPaths = getWorkerZkPaths();
int workerHeartbeatInterval =
workerConfig.getWorkerHeartbeatInterval();
-this.heartBeatExecutor.scheduleAtFixedRate(new HeartBeatTask(),
workerHeartbeatInterval, workerHeartbeatInterval, TimeUnit.SECONDS);
-logger.info("worker node : {} registry to ZK successfully with
heartBeatInterval : {}s", address, workerHeartbeatInterval);
+for (String workerZKPath : workerZkPaths) {
+
zookeeperRegistryCenter.getZookeeperCachedOperator().persistEphemeral(workerZKPath,
"");
+
zookeeperRegistryCenter.getZookeeperCachedOperator().getZkClient().getConnectionStateListenable().addListener(new
ConnectionStateListener() {
+@Override
+public void stateChanged(CuratorFramework client,
ConnectionState newState) {
+if (newState == ConnectionState.LOST) {
+logger.error("worker : {} connection lost from
zookeeper", address);
+} else if (newState == ConnectionState.RECONNECTED) {
+logger.info("worker : {} reconnected to zookeeper",
address);
+
zookeeperRegistryCenter.getZookeeperCachedOperator().persistEphemeral(workerZKPath,
"");
+} else if (newState == ConnectionState.SUSPENDED) {
+logger.warn("worker : {} connection SUSPENDED ",
address);
+}
Review comment:
> This is my suggestion, what do you think,
Thx a lot for your review.
I got your point, and can you describe your opinion about what is the
benefits of using
`if (a) { xxx; return;} if (b) { xxx; return; }` but not `if (a) { xxx; }
else if (b) {xxx;}` ?
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
[GitHub] [incubator-dolphinscheduler] yangyichao-mango commented on a change in pull request #2934: [Feature-2815][server] One worker can belong to different workergroups
yangyichao-mango commented on a change in pull request #2934:
URL:
https://github.com/apache/incubator-dolphinscheduler/pull/2934#discussion_r446907657
##
File path:
dolphinscheduler-server/src/main/java/org/apache/dolphinscheduler/server/worker/registry/WorkerRegistry.java
##
@@ -86,34 +90,39 @@ public void init(){
*/
public void registry() {
String address = OSUtils.getHost();
-String localNodePath = getWorkerPath();
-
zookeeperRegistryCenter.getZookeeperCachedOperator().persistEphemeral(localNodePath,
"");
-
zookeeperRegistryCenter.getZookeeperCachedOperator().getZkClient().getConnectionStateListenable().addListener(new
ConnectionStateListener() {
-@Override
-public void stateChanged(CuratorFramework client, ConnectionState
newState) {
-if(newState == ConnectionState.LOST){
-logger.error("worker : {} connection lost from zookeeper",
address);
-} else if(newState == ConnectionState.RECONNECTED){
-logger.info("worker : {} reconnected to zookeeper",
address);
-
zookeeperRegistryCenter.getZookeeperCachedOperator().persistEphemeral(localNodePath,
"");
-} else if(newState == ConnectionState.SUSPENDED){
-logger.warn("worker : {} connection SUSPENDED ", address);
-}
-}
-});
+Set workerZkPaths = getWorkerZkPaths();
int workerHeartbeatInterval =
workerConfig.getWorkerHeartbeatInterval();
-this.heartBeatExecutor.scheduleAtFixedRate(new HeartBeatTask(),
workerHeartbeatInterval, workerHeartbeatInterval, TimeUnit.SECONDS);
-logger.info("worker node : {} registry to ZK successfully with
heartBeatInterval : {}s", address, workerHeartbeatInterval);
+for (String workerZKPath : workerZkPaths) {
+
zookeeperRegistryCenter.getZookeeperCachedOperator().persistEphemeral(workerZKPath,
"");
+
zookeeperRegistryCenter.getZookeeperCachedOperator().getZkClient().getConnectionStateListenable().addListener(new
ConnectionStateListener() {
+@Override
+public void stateChanged(CuratorFramework client,
ConnectionState newState) {
+if (newState == ConnectionState.LOST) {
+logger.error("worker : {} connection lost from
zookeeper", address);
+} else if (newState == ConnectionState.RECONNECTED) {
+logger.info("worker : {} reconnected to zookeeper",
address);
+
zookeeperRegistryCenter.getZookeeperCachedOperator().persistEphemeral(workerZKPath,
"");
+} else if (newState == ConnectionState.SUSPENDED) {
+logger.warn("worker : {} connection SUSPENDED ",
address);
+}
Review comment:
> Hello, is it possible to return directly here without continuing to
judge? In addition, it is not very intuitive to flood with too many if else
Hi, Thx a lot for your review.
Can you describe your opinion about why it need return directly? I think
this is an important step for users to debug zk or worker status errors.
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
