[
https://issues.apache.org/jira/browse/GOBBLIN-1016?focusedWorklogId=371800&page=com.atlassian.jira.plugin.system.issuetabpanels:worklog-tabpanel#worklog-371800
]
ASF GitHub Bot logged work on GOBBLIN-1016:
-------------------------------------------
Author: ASF GitHub Bot
Created on: 14/Jan/20 19:06
Start Date: 14/Jan/20 19:06
Worklog Time Spent: 10m
Work Description: htran1 commented on pull request #2863: GOBBLIN-1016:
Allow Gobblin Application Master to join Helix cluster …
URL: https://github.com/apache/incubator-gobblin/pull/2863#discussion_r366518825
##########
File path:
gobblin-yarn/src/test/java/org/apache/gobblin/yarn/GobblinYarnAppLauncherTest.java
##########
@@ -265,17 +297,61 @@ public void testSendShutdownRequest() throws Exception {
Assert.assertEquals(this.curatorFramework.checkExists()
.forPath(String.format("/%s/CONTROLLER/MESSAGES",
GobblinYarnAppLauncherTest.class.getSimpleName()))
.getVersion(), 0);
- YarnSecurityManagerTest.GetControllerMessageNumFunc getCtrlMessageNum =
- new
YarnSecurityManagerTest.GetControllerMessageNumFunc(GobblinYarnAppLauncherTest.class.getSimpleName(),
+ YarnSecurityManagerTest.GetHelixMessageNumFunc getCtrlMessageNum =
+ new
YarnSecurityManagerTest.GetHelixMessageNumFunc(GobblinYarnAppLauncherTest.class.getSimpleName(),
InstanceType.CONTROLLER, "",
this.curatorFramework);
AssertWithBackoff assertWithBackoff =
AssertWithBackoff.create().logger(LoggerFactory.getLogger("testSendShutdownRequest")).timeoutMs(20000);
assertWithBackoff.assertEquals(getCtrlMessageNum, 1, "1 controller message
queued");
// Give Helix sometime to handle the message
assertWithBackoff.assertEquals(getCtrlMessageNum, 0, "all controller
messages processed");
+
+ this.helixManagerManagedHelix.connect();
+
this.helixManagerManagedHelix.getMessagingService().registerMessageHandlerFactory(GobblinHelixConstants.SHUTDOWN_MESSAGE_TYPE,
+ new TestShutdownMessageHandlerFactory(this));
+
+ this.gobblinYarnAppLauncherManagedHelix.connectHelixManager();
+ this.gobblinYarnAppLauncherManagedHelix.sendShutdownRequest();
+
+ Assert.assertEquals(this.curatorFramework.checkExists()
+ .forPath(String.format("/%s/INSTANCES/%s/MESSAGES",
this.configManagedHelix.getString(GobblinClusterConfigurationKeys.HELIX_CLUSTER_NAME_KEY),
TEST_HELIX_INSTANCE_NAME_MANAGED))
+ .getVersion(), 0);
+ YarnSecurityManagerTest.GetHelixMessageNumFunc getInstanceMessageNum =
+ new
YarnSecurityManagerTest.GetHelixMessageNumFunc(this.configManagedHelix.getString(
+ GobblinClusterConfigurationKeys.HELIX_CLUSTER_NAME_KEY),
+ InstanceType.PARTICIPANT, TEST_HELIX_INSTANCE_NAME_MANAGED,
this.curatorFramework);
+ assertWithBackoff =
+
AssertWithBackoff.create().logger(LoggerFactory.getLogger("testSendShutdownRequest")).timeoutMs(20000);
+ assertWithBackoff.assertEquals(getInstanceMessageNum, 1, "1 controller
message queued");
+
+ // Give Helix sometime to handle the message
+ assertWithBackoff.assertEquals(getInstanceMessageNum, 0, "all controller
messages processed");
}
+ /*static class GetInstanceMessageFunc implements Function<Void, Integer> {
Review comment:
Remove this block?
----------------------------------------------------------------
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
Issue Time Tracking
-------------------
Worklog Id: (was: 371800)
Time Spent: 50m (was: 40m)
> Allow Gobblin Application Master to join Helix cluster in PARTICIPANT mode
> when Helix cluster is managed
> --------------------------------------------------------------------------------------------------------
>
> Key: GOBBLIN-1016
> URL: https://issues.apache.org/jira/browse/GOBBLIN-1016
> Project: Apache Gobblin
> Issue Type: Improvement
> Components: gobblin-cluster
> Affects Versions: 0.15.0
> Reporter: Sudarshan Vasudevan
> Assignee: Hung Tran
> Priority: Major
> Fix For: 0.15.0
>
> Time Spent: 50m
> Remaining Estimate: 0h
>
> In the managed cluster mode, the Gobblin Yarn Application Master joins Helix
> cluster in the ADMINISTRATOR role. However, Helix does not support messaging
> for instances which are not CONTROLLER or PARTICIPANT. To properly handle
> shutdown and token file update messages, the AppMaster should connect to the
> Helix cluster as a PARTICIPANT, these message types need to be destined to
> PARTICIPANT instead of CONTROLLER as is the case today.
--
This message was sent by Atlassian Jira
(v8.3.4#803005)
