protected void writeAuditLog(ApplicationId appId) {
RMApp app = rmContext.getRMApps().get(appId);
String operation = "UNKONWN";
boolean success = false;
switch (app.getState()) {
case FAILED:
operation = AuditConstants.FINISH_FAILED_APP;
break;
case FINISHED:
operation = AuditConstants.FINISH_SUCCESS_APP;
success = true;
break;
case KILLED:
operation = AuditConstants.FINISH_KILLED_APP;
success = true;
break;
default:
}
if (success) {
RMAuditLogger.logSuccess(app.getUser(), operation, "RMAppManager", app.getApplicationId());
} else {
StringBuilder diag = app.getDiagnostics();
String msg = diag == null ? null : diag.toString();
RMAuditLogger.logFailure(
app.getUser(),
operation,
msg,
"RMAppManager",
"App failed with state: " + app.getState(),
appId);
}
}
@Override
public RefreshUserToGroupsMappingsResponse refreshUserToGroupsMappings(
RefreshUserToGroupsMappingsRequest request) throws YarnException, IOException {
String argName = "refreshUserToGroupsMappings";
UserGroupInformation user = checkAcls(argName);
if (!isRMActive()) {
RMAuditLogger.logFailure(
user.getShortUserName(),
argName,
adminAcl.toString(),
"AdminService",
"ResourceManager is not active. Can not refresh user-groups.");
throwStandbyException();
}
Groups.getUserToGroupsMappingService(
getConfiguration(
new Configuration(false), YarnConfiguration.CORE_SITE_CONFIGURATION_FILE))
.refresh();
RMAuditLogger.logSuccess(user.getShortUserName(), argName, "AdminService");
return recordFactory.newRecordInstance(RefreshUserToGroupsMappingsResponse.class);
}
@Override
public RefreshNodesResponse refreshNodes(RefreshNodesRequest request)
throws YarnException, StandbyException {
String argName = "refreshNodes";
UserGroupInformation user = checkAcls("refreshNodes");
if (!isRMActive()) {
RMAuditLogger.logFailure(
user.getShortUserName(),
argName,
adminAcl.toString(),
"AdminService",
"ResourceManager is not active. Can not refresh nodes.");
throwStandbyException();
}
try {
Configuration conf =
getConfiguration(
new Configuration(false), YarnConfiguration.YARN_SITE_CONFIGURATION_FILE);
rmContext.getNodesListManager().refreshNodes(conf);
RMAuditLogger.logSuccess(user.getShortUserName(), argName, "AdminService");
return recordFactory.newRecordInstance(RefreshNodesResponse.class);
} catch (IOException ioe) {
LOG.info("Exception refreshing nodes ", ioe);
RMAuditLogger.logFailure(
user.getShortUserName(),
argName,
adminAcl.toString(),
"AdminService",
"Exception refreshing nodes");
throw RPCUtil.getRemoteException(ioe);
}
}
@Override
public RefreshQueuesResponse refreshQueues(RefreshQueuesRequest request)
throws YarnException, StandbyException {
String argName = "refreshQueues";
UserGroupInformation user = checkAcls(argName);
if (!isRMActive()) {
RMAuditLogger.logFailure(
user.getShortUserName(),
argName,
adminAcl.toString(),
"AdminService",
"ResourceManager is not active. Can not refresh queues.");
throwStandbyException();
}
RefreshQueuesResponse response = recordFactory.newRecordInstance(RefreshQueuesResponse.class);
try {
rmContext.getScheduler().reinitialize(getConfig(), this.rmContext);
RMAuditLogger.logSuccess(user.getShortUserName(), argName, "AdminService");
return response;
} catch (IOException ioe) {
LOG.info("Exception refreshing queues ", ioe);
RMAuditLogger.logFailure(
user.getShortUserName(),
argName,
adminAcl.toString(),
"AdminService",
"Exception refreshing queues");
throw RPCUtil.getRemoteException(ioe);
}
}
@Override
public synchronized void transitionToStandby(HAServiceProtocol.StateChangeRequestInfo reqInfo)
throws IOException {
// call refreshAdminAcls before HA state transition
// for the case that adminAcls have been updated in previous active RM
try {
refreshAdminAcls(false);
} catch (YarnException ex) {
throw new ServiceFailedException("Can not execute refreshAdminAcls", ex);
}
UserGroupInformation user = checkAccess("transitionToStandby");
checkHaStateChange(reqInfo);
try {
LOG.info("Transitioning to standby admin" + masterServiceAddress.toString());
// TODO transition leader election to standby?
rm.transitionToStandby(true);
RMAuditLogger.logSuccess(
user.getShortUserName(), "transitionToStandby", "RMHAProtocolService");
} catch (Exception e) {
RMAuditLogger.logFailure(
user.getShortUserName(),
"transitionToStandby",
adminAcl.toString(),
"RMHAProtocolService",
"Exception transitioning to standby");
throw new ServiceFailedException("Error when transitioning to Standby mode", e);
}
}
@Override
public synchronized void transitionToActive(HAServiceProtocol.StateChangeRequestInfo reqInfo)
throws IOException {
// call refreshAdminAcls before HA state transition
// for the case that adminAcls have been updated in previous active RM
try {
refreshAdminAcls(false);
} catch (YarnException ex) {
throw new ServiceFailedException("Can not execute refreshAdminAcls", ex);
}
UserGroupInformation user = checkAccess("transitionToActive");
checkHaStateChange(reqInfo);
try {
if (!autoFailoverEnabled) {
LOG.info("admin transition to active " + masterServiceAddress.toString());
rm.transitionToActive();
// call all refresh*s for active RM to get the updated configurations.
refreshAll();
RMAuditLogger.logSuccess(
user.getShortUserName(), "transitionToActive", "RMHAProtocolService");
}
} catch (Exception e) {
RMAuditLogger.logFailure(
user.getShortUserName(),
"transitionToActive",
adminAcl.toString(),
"RMHAProtocolService",
"Exception transitioning to active");
throw new ServiceFailedException("Error when transitioning to Active mode", e);
}
}
public synchronized void containerCompleted(
RMContainer rmContainer, ContainerStatus containerStatus, RMContainerEventType event) {
Container container = rmContainer.getContainer();
ContainerId containerId = container.getId();
// Remove from the list of newly allocated containers if found
newlyAllocatedContainers.remove(rmContainer);
// Inform the container
rmContainer.handle(new RMContainerFinishedEvent(containerId, containerStatus, event));
LOG.info(
"Completed container: "
+ rmContainer.getContainerId()
+ " in state: "
+ rmContainer.getState()
+ " event:"
+ event);
// Remove from the list of containers
liveContainers.remove(rmContainer.getContainerId());
RMAuditLogger.logSuccess(
getUser(),
AuditConstants.RELEASE_CONTAINER,
"SchedulerApp",
getApplicationId(),
containerId);
// Update usage metrics
Resource containerResource = rmContainer.getContainer().getResource();
queue.getMetrics().releaseResources(getUser(), 1, containerResource);
Resources.subtractFrom(currentConsumption, containerResource);
// remove from preemption map if it is completed
preemptionMap.remove(rmContainer);
// Clear resource utilization metrics cache.
lastMemoryAggregateAllocationUpdateTime = -1;
}
private RefreshAdminAclsResponse refreshAdminAcls(boolean checkRMHAState)
throws YarnException, IOException {
String argName = "refreshAdminAcls";
UserGroupInformation user = checkAcls(argName);
if (checkRMHAState && !isRMActive()) {
RMAuditLogger.logFailure(
user.getShortUserName(),
argName,
adminAcl.toString(),
"AdminService",
"ResourceManager is not active. Can not refresh user-groups.");
throwStandbyException();
}
Configuration conf =
getConfiguration(new Configuration(false), YarnConfiguration.YARN_SITE_CONFIGURATION_FILE);
adminAcl =
new AccessControlList(
conf.get(YarnConfiguration.YARN_ADMIN_ACL, YarnConfiguration.DEFAULT_YARN_ADMIN_ACL));
RMAuditLogger.logSuccess(user.getShortUserName(), argName, "AdminService");
return recordFactory.newRecordInstance(RefreshAdminAclsResponse.class);
}
public synchronized RMContainer allocate(
NodeType type,
FSSchedulerNode node,
Priority priority,
ResourceRequest request,
Container container) {
// Update allowed locality level
NodeType allowed = allowedLocalityLevel.get(priority);
if (allowed != null) {
if (allowed.equals(NodeType.OFF_SWITCH)
&& (type.equals(NodeType.NODE_LOCAL) || type.equals(NodeType.RACK_LOCAL))) {
this.resetAllowedLocalityLevel(priority, type);
} else if (allowed.equals(NodeType.RACK_LOCAL) && type.equals(NodeType.NODE_LOCAL)) {
this.resetAllowedLocalityLevel(priority, type);
}
}
// Required sanity check - AM can call 'allocate' to update resource
// request without locking the scheduler, hence we need to check
if (getTotalRequiredResources(priority) <= 0) {
return null;
}
// Create RMContainer
RMContainer rmContainer =
new RMContainerImpl(
container,
getApplicationAttemptId(),
node.getNodeID(),
appSchedulingInfo.getUser(),
rmContext);
// Add it to allContainers list.
newlyAllocatedContainers.add(rmContainer);
liveContainers.put(container.getId(), rmContainer);
// Update consumption and track allocations
List<ResourceRequest> resourceRequestList =
appSchedulingInfo.allocate(type, node, priority, request, container);
Resources.addTo(currentConsumption, container.getResource());
// Update resource requests related to "request" and store in RMContainer
((RMContainerImpl) rmContainer).setResourceRequests(resourceRequestList);
// Inform the container
rmContainer.handle(new RMContainerEvent(container.getId(), RMContainerEventType.START));
if (LOG.isDebugEnabled()) {
LOG.debug(
"allocate: applicationAttemptId="
+ container.getId().getApplicationAttemptId()
+ " container="
+ container.getId()
+ " host="
+ container.getNodeId().getHost()
+ " type="
+ type);
}
RMAuditLogger.logSuccess(
getUser(),
AuditConstants.ALLOC_CONTAINER,
"SchedulerApp",
getApplicationId(),
container.getId());
return rmContainer;
}