Merging trunk to HDFS-2802 branch.

git-svn-id: https://svn.apache.org/repos/asf/hadoop/common/branches/HDFS-2802@1404629 13f79535-47bb-0310-9956-ffa450edef68
This commit is contained in:
Suresh Srinivas 2012-11-01 15:11:27 +00:00
commit 04b2adca05
4 changed files with 25 additions and 12 deletions

View File

@ -620,6 +620,9 @@ Release 0.23.5 - UNRELEASED
MAPREDUCE-4748. Invalid event: T_ATTEMPT_SUCCEEDED at SUCCEEDED. (jlowe)
MAPREDUCE-4724. job history web ui applications page should be sorted to
display last app first (tgraves via bobby)
Release 0.23.4 - UNRELEASED
INCOMPATIBLE CHANGES

View File

@ -85,7 +85,7 @@ public class HsView extends TwoColumnLayout {
private String jobsTableInit() {
return tableInit().
// Sort by id upon page load
append(", aaSorting: [[2, 'asc']]").
append(", aaSorting: [[2, 'desc']]").
append(", aoColumnDefs:[").
// Maps Total
append("{'sType':'numeric', 'bSearchable': false, 'aTargets': [ 7 ] }").

View File

@ -196,6 +196,9 @@ Release 0.23.5 - UNRELEASED
YARN-166. capacity scheduler doesn't allow capacity < 1.0 (tgraves via
bobby)
YARN-189. Fixed a deadlock between RM's ApplicationMasterService and the
dispatcher. (Thomas Graves via vinodkv)
Release 0.23.4 - UNRELEASED
INCOMPATIBLE CHANGES

View File

@ -265,10 +265,10 @@ public class ApplicationMasterService extends AbstractService implements
// Oh damn! Sending reboot isn't enough. RM state is corrupted. TODO:
allocateResponse.setAMResponse(reboot);
return allocateResponse;
}
}
// Allow only one thread in AM to do heartbeat at a time.
synchronized (lastResponse) { // BUG TODO: Locking order is screwed.
synchronized (lastResponse) {
// Send the status update to the appAttempt.
this.rmContext.getDispatcher().getEventHandler().handle(
@ -282,7 +282,8 @@ public class ApplicationMasterService extends AbstractService implements
Allocation allocation =
this.rScheduler.allocate(appAttemptId, ask, release);
RMApp app = this.rmContext.getRMApps().get(appAttemptId.getApplicationId());
RMApp app = this.rmContext.getRMApps().get(
appAttemptId.getApplicationId());
RMAppAttempt appAttempt = app.getRMAppAttempt(appAttemptId);
AMResponse response = recordFactory.newRecordInstance(AMResponse.class);
@ -316,7 +317,18 @@ public class ApplicationMasterService extends AbstractService implements
.pullJustFinishedContainers());
response.setResponseId(lastResponse.getResponseId() + 1);
response.setAvailableResources(allocation.getResourceLimit());
responseMap.put(appAttemptId, response);
AMResponse oldResponse = responseMap.put(appAttemptId, response);
if (oldResponse == null) {
// appAttempt got unregistered, remove it back out
responseMap.remove(appAttemptId);
String message = "App Attempt removed from the cache during allocate"
+ appAttemptId;
LOG.error(message);
allocateResponse.setAMResponse(reboot);
return allocateResponse;
}
allocateResponse.setAMResponse(response);
allocateResponse.setNumClusterNodes(this.rScheduler.getNumClusterNodes());
return allocateResponse;
@ -331,12 +343,7 @@ public class ApplicationMasterService extends AbstractService implements
}
public void unregisterAttempt(ApplicationAttemptId attemptId) {
AMResponse lastResponse = responseMap.get(attemptId);
if (lastResponse != null) {
synchronized (lastResponse) {
responseMap.remove(attemptId);
}
}
responseMap.remove(attemptId);
}
public void refreshServiceAcls(Configuration configuration,