Apache
/
hadoop
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

YARN-876. Node resource is added twice when node comes back from unhealthy. (Peng Zhang via Sandy Ryza) 

git-svn-id: https://svn.apache.org/repos/asf/hadoop/common/branches/branch-2@1528661 13f79535-47bb-0310-9956-ffa450edef68
This commit is contained in:
Sanford Ryza 2013-10-02 22:35:37 +00:00
parent c738c106e9
commit 9ba48807df
3 changed files with 16 additions and 2 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

3
hadoop-yarn-project/CHANGES.txt
View File

@ -114,6 +114,9 @@ Release 2.1.2 - UNRELEASED
YARN-1141. Updating resource requests should be decoupled with updating
blacklist (Zhijie Shen via bikas)
YARN-876. Node resource is added twice when node comes back from unhealthy
to healthy. (Peng Zhang via Sandy Ryza)
Release 2.1.1-beta - 2013-09-23
INCOMPATIBLE CHANGES

7
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/main/java/org/apache/hadoop/yarn/server/resourcemanager/rmnode/RMNodeImpl.java
View File

@ -460,8 +460,11 @@ public class RMNodeImpl implements RMNode, EventHandler<RMNodeEvent> {
&& rmNode.getHttpPort() == newNode.getHttpPort()) {
// Reset heartbeat ID since node just restarted.
rmNode.getLastNodeHeartBeatResponse().setResponseId(0);
rmNode.context.getDispatcher().getEventHandler().handle(
new NodeAddedSchedulerEvent(rmNode));
if (rmNode.getState() != NodeState.UNHEALTHY) {
// Only add new node if old state is not UNHEALTHY
rmNode.context.getDispatcher().getEventHandler().handle(
new NodeAddedSchedulerEvent(rmNode));
}
} else {
// Reconnected node differs, so replace old node and start new node
switch (rmNode.getState()) {

8
hadoop-yarn-project/hadoop-yarn/hadoop-yarn-server/hadoop-yarn-server-resourcemanager/src/test/java/org/apache/hadoop/yarn/server/resourcemanager/TestResourceTrackerService.java
View File

@ -506,6 +506,14 @@ public class TestResourceTrackerService {
dispatcher.await();
Assert.assertEquals(expectedNMs, ClusterMetrics.getMetrics().getNumActiveNMs());
checkUnealthyNMCount(rm, nm2, true, 1);
// unhealthy node changed back to healthy
nm2 = rm.registerNode("host2:5678", 5120);
dispatcher.await();
response = nm2.nodeHeartbeat(true);
response = nm2.nodeHeartbeat(true);
dispatcher.await();
Assert.assertEquals(5120 + 5120, metrics.getAvailableMB());
// reconnect of node with changed capability
nm1 = rm.registerNode("host2:5678", 10240);