YARN-6475. Fix some long function checkstyle issues
(Contributed by Soumabrata Chakraborty via Daniel Templeton)
This commit is contained in:
Daniel Templeton
parent
1e71fe8c42
commit
74a61438ca
@ -62,7 +62,6 @@
|
||||
import java.util.ArrayList;
|
||||
import java.util.Arrays;
|
||||
import java.util.List;
|
||||
import java.util.Map;
|
||||
import java.util.regex.Pattern;
|
||||
|
||||
import static org.apache.hadoop.yarn.server.nodemanager.containermanager.linux.runtime.LinuxContainerRuntimeConstants.*;
|
||||
@ -442,24 +441,11 @@ public void prepareContainer(ContainerPrepareContext ctx) throws IOException {
|
||||
public int launchContainer(ContainerStartContext ctx)
|
||||
throws IOException, ConfigurationException {
|
||||
Container container = ctx.getContainer();
|
||||
Path nmPrivateContainerScriptPath = ctx.getNmPrivateContainerScriptPath();
|
||||
Path nmPrivateTokensPath = ctx.getNmPrivateTokensPath();
|
||||
String user = ctx.getUser();
|
||||
String appId = ctx.getAppId();
|
||||
Path containerWorkDir = ctx.getContainerWorkDir();
|
||||
List<String> localDirs = ctx.getLocalDirs();
|
||||
List<String> logDirs = ctx.getLogDirs();
|
||||
List<String> filecacheDirs = ctx.getFilecacheDirs();
|
||||
List<String> userLocalDirs = ctx.getUserLocalDirs();
|
||||
List<String> containerLocalDirs = ctx.getContainerLocalDirs();
|
||||
List<String> containerLogDirs = ctx.getContainerLogDirs();
|
||||
Map<Path, List<String>> localizedResources = ctx.getLocalizedResources();
|
||||
|
||||
verifyUsernamePattern(user);
|
||||
String runAsUser = getRunAsUser(user);
|
||||
|
||||
ContainerId containerId = container.getContainerId();
|
||||
String containerIdStr = containerId.toString();
|
||||
|
||||
resourcesHandler.preExecute(containerId,
|
||||
container.getResource());
|
||||
@ -514,39 +500,11 @@ public int launchContainer(ContainerStartContext ctx)
|
||||
try {
|
||||
Path pidFilePath = getPidFilePath(containerId);
|
||||
if (pidFilePath != null) {
|
||||
List<String> prefixCommands = new ArrayList<>();
|
||||
ContainerRuntimeContext.Builder builder = new ContainerRuntimeContext
|
||||
.Builder(container);
|
||||
|
||||
addSchedPriorityCommand(prefixCommands);
|
||||
if (prefixCommands.size() > 0) {
|
||||
builder.setExecutionAttribute(CONTAINER_LAUNCH_PREFIX_COMMANDS,
|
||||
prefixCommands);
|
||||
}
|
||||
ContainerRuntimeContext runtimeContext = buildContainerRuntimeContext(
|
||||
ctx, pidFilePath, resourcesOptions, tcCommandFile);
|
||||
|
||||
builder.setExecutionAttribute(LOCALIZED_RESOURCES, localizedResources)
|
||||
.setExecutionAttribute(RUN_AS_USER, runAsUser)
|
||||
.setExecutionAttribute(USER, user)
|
||||
.setExecutionAttribute(APPID, appId)
|
||||
.setExecutionAttribute(CONTAINER_ID_STR, containerIdStr)
|
||||
.setExecutionAttribute(CONTAINER_WORK_DIR, containerWorkDir)
|
||||
.setExecutionAttribute(NM_PRIVATE_CONTAINER_SCRIPT_PATH,
|
||||
nmPrivateContainerScriptPath)
|
||||
.setExecutionAttribute(NM_PRIVATE_TOKENS_PATH, nmPrivateTokensPath)
|
||||
.setExecutionAttribute(PID_FILE_PATH, pidFilePath)
|
||||
.setExecutionAttribute(LOCAL_DIRS, localDirs)
|
||||
.setExecutionAttribute(LOG_DIRS, logDirs)
|
||||
.setExecutionAttribute(FILECACHE_DIRS, filecacheDirs)
|
||||
.setExecutionAttribute(USER_LOCAL_DIRS, userLocalDirs)
|
||||
.setExecutionAttribute(CONTAINER_LOCAL_DIRS, containerLocalDirs)
|
||||
.setExecutionAttribute(CONTAINER_LOG_DIRS, containerLogDirs)
|
||||
.setExecutionAttribute(RESOURCES_OPTIONS, resourcesOptions);
|
||||
|
||||
if (tcCommandFile != null) {
|
||||
builder.setExecutionAttribute(TC_COMMAND_FILE, tcCommandFile);
|
||||
}
|
||||
|
||||
linuxContainerRuntime.launchContainer(builder.build());
|
||||
linuxContainerRuntime.launchContainer(runtimeContext);
|
||||
} else {
|
||||
LOG.info(
|
||||
"Container was marked as inactive. Returning terminated error");
|
||||
@ -617,6 +575,50 @@ public int launchContainer(ContainerStartContext ctx)
|
||||
return 0;
|
||||
}
|
||||
|
||||
private ContainerRuntimeContext buildContainerRuntimeContext(
|
||||
ContainerStartContext ctx, Path pidFilePath,
|
||||
String resourcesOptions, String tcCommandFile) {
|
||||
|
||||
List<String> prefixCommands = new ArrayList<>();
|
||||
addSchedPriorityCommand(prefixCommands);
|
||||
|
||||
Container container = ctx.getContainer();
|
||||
|
||||
ContainerRuntimeContext.Builder builder = new ContainerRuntimeContext
|
||||
.Builder(container);
|
||||
if (prefixCommands.size() > 0) {
|
||||
builder.setExecutionAttribute(CONTAINER_LAUNCH_PREFIX_COMMANDS,
|
||||
prefixCommands);
|
||||
}
|
||||
|
||||
builder.setExecutionAttribute(LOCALIZED_RESOURCES,
|
||||
ctx.getLocalizedResources())
|
||||
.setExecutionAttribute(RUN_AS_USER, getRunAsUser(ctx.getUser()))
|
||||
.setExecutionAttribute(USER, ctx.getUser())
|
||||
.setExecutionAttribute(APPID, ctx.getAppId())
|
||||
.setExecutionAttribute(CONTAINER_ID_STR,
|
||||
container.getContainerId().toString())
|
||||
.setExecutionAttribute(CONTAINER_WORK_DIR, ctx.getContainerWorkDir())
|
||||
.setExecutionAttribute(NM_PRIVATE_CONTAINER_SCRIPT_PATH,
|
||||
ctx.getNmPrivateContainerScriptPath())
|
||||
.setExecutionAttribute(NM_PRIVATE_TOKENS_PATH,
|
||||
ctx.getNmPrivateTokensPath())
|
||||
.setExecutionAttribute(PID_FILE_PATH, pidFilePath)
|
||||
.setExecutionAttribute(LOCAL_DIRS, ctx.getLocalDirs())
|
||||
.setExecutionAttribute(LOG_DIRS, ctx.getLogDirs())
|
||||
.setExecutionAttribute(FILECACHE_DIRS, ctx.getFilecacheDirs())
|
||||
.setExecutionAttribute(USER_LOCAL_DIRS, ctx.getUserLocalDirs())
|
||||
.setExecutionAttribute(CONTAINER_LOCAL_DIRS, ctx.getContainerLocalDirs())
|
||||
.setExecutionAttribute(CONTAINER_LOG_DIRS, ctx.getContainerLogDirs())
|
||||
.setExecutionAttribute(RESOURCES_OPTIONS, resourcesOptions);
|
||||
|
||||
if (tcCommandFile != null) {
|
||||
builder.setExecutionAttribute(TC_COMMAND_FILE, tcCommandFile);
|
||||
}
|
||||
|
||||
return builder.build();
|
||||
}
|
||||
|
||||
@Override
|
||||
public String[] getIpAndHost(Container container) {
|
||||
return linuxContainerRuntime.getIpAndHost(container);
|
||||
|
||||
@ -761,200 +761,7 @@ private static Map<ApplicationId, Credentials> parseCredentials(
|
||||
|
||||
protected void startStatusUpdater() {
|
||||
|
||||
statusUpdaterRunnable = new Runnable() {
|
||||
@Override
|
||||
@SuppressWarnings("unchecked")
|
||||
public void run() {
|
||||
int lastHeartbeatID = 0;
|
||||
while (!isStopped) {
|
||||
// Send heartbeat
|
||||
try {
|
||||
NodeHeartbeatResponse response = null;
|
||||
Set<NodeLabel> nodeLabelsForHeartbeat =
|
||||
nodeLabelsHandler.getNodeLabelsForHeartbeat();
|
||||
NodeStatus nodeStatus = getNodeStatus(lastHeartbeatID);
|
||||
NodeHeartbeatRequest request =
|
||||
NodeHeartbeatRequest.newInstance(nodeStatus,
|
||||
NodeStatusUpdaterImpl.this.context
|
||||
.getContainerTokenSecretManager().getCurrentKey(),
|
||||
NodeStatusUpdaterImpl.this.context
|
||||
.getNMTokenSecretManager().getCurrentKey(),
|
||||
nodeLabelsForHeartbeat,
|
||||
NodeStatusUpdaterImpl.this.context
|
||||
.getRegisteredCollectors());
|
||||
|
||||
if (logAggregationEnabled) {
|
||||
// pull log aggregation status for application running in this NM
|
||||
List<LogAggregationReport> logAggregationReports =
|
||||
getLogAggregationReportsForApps(context
|
||||
.getLogAggregationStatusForApps());
|
||||
if (logAggregationReports != null
|
||||
&& !logAggregationReports.isEmpty()) {
|
||||
request.setLogAggregationReportsForApps(logAggregationReports);
|
||||
}
|
||||
}
|
||||
|
||||
response = resourceTracker.nodeHeartbeat(request);
|
||||
//get next heartbeat interval from response
|
||||
nextHeartBeatInterval = response.getNextHeartBeatInterval();
|
||||
updateMasterKeys(response);
|
||||
|
||||
if (!handleShutdownOrResyncCommand(response)) {
|
||||
nodeLabelsHandler.verifyRMHeartbeatResponseForNodeLabels(
|
||||
response);
|
||||
|
||||
// Explicitly put this method after checking the resync
|
||||
// response. We
|
||||
// don't want to remove the completed containers before resync
|
||||
// because these completed containers will be reported back to RM
|
||||
// when NM re-registers with RM.
|
||||
// Only remove the cleanedup containers that are acked
|
||||
removeOrTrackCompletedContainersFromContext(response
|
||||
.getContainersToBeRemovedFromNM());
|
||||
|
||||
logAggregationReportForAppsTempList.clear();
|
||||
lastHeartbeatID = response.getResponseId();
|
||||
List<ContainerId> containersToCleanup = response
|
||||
.getContainersToCleanup();
|
||||
if (!containersToCleanup.isEmpty()) {
|
||||
dispatcher.getEventHandler().handle(
|
||||
new CMgrCompletedContainersEvent(containersToCleanup,
|
||||
CMgrCompletedContainersEvent.Reason
|
||||
.BY_RESOURCEMANAGER));
|
||||
}
|
||||
List<ApplicationId> appsToCleanup =
|
||||
response.getApplicationsToCleanup();
|
||||
//Only start tracking for keepAlive on FINISH_APP
|
||||
trackAppsForKeepAlive(appsToCleanup);
|
||||
if (!appsToCleanup.isEmpty()) {
|
||||
dispatcher.getEventHandler().handle(
|
||||
new CMgrCompletedAppsEvent(appsToCleanup,
|
||||
CMgrCompletedAppsEvent.Reason.BY_RESOURCEMANAGER));
|
||||
}
|
||||
Map<ApplicationId, ByteBuffer> systemCredentials =
|
||||
response.getSystemCredentialsForApps();
|
||||
if (systemCredentials != null && !systemCredentials.isEmpty()) {
|
||||
((NMContext) context).setSystemCrendentialsForApps(
|
||||
parseCredentials(systemCredentials));
|
||||
}
|
||||
List<org.apache.hadoop.yarn.api.records.Container>
|
||||
containersToDecrease = response.getContainersToDecrease();
|
||||
if (!containersToDecrease.isEmpty()) {
|
||||
dispatcher.getEventHandler().handle(
|
||||
new CMgrDecreaseContainersResourceEvent(
|
||||
containersToDecrease)
|
||||
);
|
||||
}
|
||||
|
||||
// SignalContainer request originally comes from end users via
|
||||
// ClientRMProtocol's SignalContainer. Forward the request to
|
||||
// ContainerManager which will dispatch the event to
|
||||
// ContainerLauncher.
|
||||
List<SignalContainerRequest> containersToSignal = response
|
||||
.getContainersToSignalList();
|
||||
if (containersToSignal.size() != 0) {
|
||||
dispatcher.getEventHandler().handle(
|
||||
new CMgrSignalContainersEvent(containersToSignal));
|
||||
}
|
||||
|
||||
// Update QueuingLimits if ContainerManager supports queuing
|
||||
ContainerQueuingLimit queuingLimit =
|
||||
response.getContainerQueuingLimit();
|
||||
if (queuingLimit != null) {
|
||||
context.getContainerManager().updateQueuingLimit(queuingLimit);
|
||||
}
|
||||
}
|
||||
// Handling node resource update case.
|
||||
Resource newResource = response.getResource();
|
||||
if (newResource != null) {
|
||||
updateNMResource(newResource);
|
||||
if (LOG.isDebugEnabled()) {
|
||||
LOG.debug("Node's resource is updated to " +
|
||||
newResource.toString());
|
||||
}
|
||||
}
|
||||
if (YarnConfiguration.timelineServiceV2Enabled(context.getConf())) {
|
||||
updateTimelineClientsAddress(response);
|
||||
}
|
||||
|
||||
} catch (ConnectException e) {
|
||||
//catch and throw the exception if tried MAX wait time to connect RM
|
||||
dispatcher.getEventHandler().handle(
|
||||
new NodeManagerEvent(NodeManagerEventType.SHUTDOWN));
|
||||
// failed to connect to RM.
|
||||
failedToConnect = true;
|
||||
throw new YarnRuntimeException(e);
|
||||
} catch (Throwable e) {
|
||||
|
||||
// TODO Better error handling. Thread can die with the rest of the
|
||||
// NM still running.
|
||||
LOG.error("Caught exception in status-updater", e);
|
||||
} finally {
|
||||
synchronized (heartbeatMonitor) {
|
||||
nextHeartBeatInterval = nextHeartBeatInterval <= 0 ?
|
||||
YarnConfiguration.DEFAULT_RM_NM_HEARTBEAT_INTERVAL_MS :
|
||||
nextHeartBeatInterval;
|
||||
try {
|
||||
heartbeatMonitor.wait(nextHeartBeatInterval);
|
||||
} catch (InterruptedException e) {
|
||||
// Do Nothing
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
private void updateTimelineClientsAddress(
|
||||
NodeHeartbeatResponse response) {
|
||||
Map<ApplicationId, String> knownCollectorsMap =
|
||||
response.getAppCollectorsMap();
|
||||
if (knownCollectorsMap == null) {
|
||||
if (LOG.isDebugEnabled()) {
|
||||
LOG.debug("No collectors to update RM");
|
||||
}
|
||||
} else {
|
||||
Set<Map.Entry<ApplicationId, String>> rmKnownCollectors =
|
||||
knownCollectorsMap.entrySet();
|
||||
for (Map.Entry<ApplicationId, String> entry : rmKnownCollectors) {
|
||||
ApplicationId appId = entry.getKey();
|
||||
String collectorAddr = entry.getValue();
|
||||
|
||||
// Only handle applications running on local node.
|
||||
// Not include apps with timeline collectors running in local
|
||||
Application application = context.getApplications().get(appId);
|
||||
// TODO this logic could be problematic if the collector address
|
||||
// gets updated due to NM restart or collector service failure
|
||||
if (application != null &&
|
||||
!context.getRegisteredCollectors().containsKey(appId)) {
|
||||
if (LOG.isDebugEnabled()) {
|
||||
LOG.debug("Sync a new collector address: " + collectorAddr +
|
||||
" for application: " + appId + " from RM.");
|
||||
}
|
||||
NMTimelinePublisher nmTimelinePublisher =
|
||||
context.getNMTimelinePublisher();
|
||||
if (nmTimelinePublisher != null) {
|
||||
nmTimelinePublisher.setTimelineServiceAddress(
|
||||
application.getAppId(), collectorAddr);
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
private void updateMasterKeys(NodeHeartbeatResponse response) {
|
||||
// See if the master-key has rolled over
|
||||
MasterKey updatedMasterKey = response.getContainerTokenMasterKey();
|
||||
if (updatedMasterKey != null) {
|
||||
// Will be non-null only on roll-over on RM side
|
||||
context.getContainerTokenSecretManager().setMasterKey(updatedMasterKey);
|
||||
}
|
||||
|
||||
updatedMasterKey = response.getNMTokenMasterKey();
|
||||
if (updatedMasterKey != null) {
|
||||
context.getNMTokenSecretManager().setMasterKey(updatedMasterKey);
|
||||
}
|
||||
}
|
||||
};
|
||||
statusUpdaterRunnable = new StatusUpdaterRunnable();
|
||||
statusUpdater =
|
||||
new Thread(statusUpdaterRunnable, "Node Status Updater");
|
||||
statusUpdater.start();
|
||||
@ -1215,4 +1022,199 @@ public void verifyRMHeartbeatResponseForNodeLabels(
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
private class StatusUpdaterRunnable implements Runnable {
|
||||
@Override
|
||||
@SuppressWarnings("unchecked")
|
||||
public void run() {
|
||||
int lastHeartbeatID = 0;
|
||||
while (!isStopped) {
|
||||
// Send heartbeat
|
||||
try {
|
||||
NodeHeartbeatResponse response = null;
|
||||
Set<NodeLabel> nodeLabelsForHeartbeat =
|
||||
nodeLabelsHandler.getNodeLabelsForHeartbeat();
|
||||
NodeStatus nodeStatus = getNodeStatus(lastHeartbeatID);
|
||||
NodeHeartbeatRequest request =
|
||||
NodeHeartbeatRequest.newInstance(nodeStatus,
|
||||
NodeStatusUpdaterImpl.this.context
|
||||
.getContainerTokenSecretManager().getCurrentKey(),
|
||||
NodeStatusUpdaterImpl.this.context
|
||||
.getNMTokenSecretManager().getCurrentKey(),
|
||||
nodeLabelsForHeartbeat,
|
||||
NodeStatusUpdaterImpl.this.context
|
||||
.getRegisteredCollectors());
|
||||
|
||||
if (logAggregationEnabled) {
|
||||
// pull log aggregation status for application running in this NM
|
||||
List<LogAggregationReport> logAggregationReports =
|
||||
getLogAggregationReportsForApps(context
|
||||
.getLogAggregationStatusForApps());
|
||||
if (logAggregationReports != null
|
||||
&& !logAggregationReports.isEmpty()) {
|
||||
request.setLogAggregationReportsForApps(logAggregationReports);
|
||||
}
|
||||
}
|
||||
|
||||
response = resourceTracker.nodeHeartbeat(request);
|
||||
//get next heartbeat interval from response
|
||||
nextHeartBeatInterval = response.getNextHeartBeatInterval();
|
||||
updateMasterKeys(response);
|
||||
|
||||
if (!handleShutdownOrResyncCommand(response)) {
|
||||
nodeLabelsHandler.verifyRMHeartbeatResponseForNodeLabels(
|
||||
response);
|
||||
|
||||
// Explicitly put this method after checking the resync
|
||||
// response. We
|
||||
// don't want to remove the completed containers before resync
|
||||
// because these completed containers will be reported back to RM
|
||||
// when NM re-registers with RM.
|
||||
// Only remove the cleanedup containers that are acked
|
||||
removeOrTrackCompletedContainersFromContext(response
|
||||
.getContainersToBeRemovedFromNM());
|
||||
|
||||
logAggregationReportForAppsTempList.clear();
|
||||
lastHeartbeatID = response.getResponseId();
|
||||
List<ContainerId> containersToCleanup = response
|
||||
.getContainersToCleanup();
|
||||
if (!containersToCleanup.isEmpty()) {
|
||||
dispatcher.getEventHandler().handle(
|
||||
new CMgrCompletedContainersEvent(containersToCleanup,
|
||||
CMgrCompletedContainersEvent.Reason
|
||||
.BY_RESOURCEMANAGER));
|
||||
}
|
||||
List<ApplicationId> appsToCleanup =
|
||||
response.getApplicationsToCleanup();
|
||||
//Only start tracking for keepAlive on FINISH_APP
|
||||
trackAppsForKeepAlive(appsToCleanup);
|
||||
if (!appsToCleanup.isEmpty()) {
|
||||
dispatcher.getEventHandler().handle(
|
||||
new CMgrCompletedAppsEvent(appsToCleanup,
|
||||
CMgrCompletedAppsEvent.Reason.BY_RESOURCEMANAGER));
|
||||
}
|
||||
Map<ApplicationId, ByteBuffer> systemCredentials =
|
||||
response.getSystemCredentialsForApps();
|
||||
if (systemCredentials != null && !systemCredentials.isEmpty()) {
|
||||
((NMContext) context).setSystemCrendentialsForApps(
|
||||
parseCredentials(systemCredentials));
|
||||
}
|
||||
List<org.apache.hadoop.yarn.api.records.Container>
|
||||
containersToDecrease = response.getContainersToDecrease();
|
||||
if (!containersToDecrease.isEmpty()) {
|
||||
dispatcher.getEventHandler().handle(
|
||||
new CMgrDecreaseContainersResourceEvent(
|
||||
containersToDecrease)
|
||||
);
|
||||
}
|
||||
|
||||
// SignalContainer request originally comes from end users via
|
||||
// ClientRMProtocol's SignalContainer. Forward the request to
|
||||
// ContainerManager which will dispatch the event to
|
||||
// ContainerLauncher.
|
||||
List<SignalContainerRequest> containersToSignal = response
|
||||
.getContainersToSignalList();
|
||||
if (!containersToSignal.isEmpty()) {
|
||||
dispatcher.getEventHandler().handle(
|
||||
new CMgrSignalContainersEvent(containersToSignal));
|
||||
}
|
||||
|
||||
// Update QueuingLimits if ContainerManager supports queuing
|
||||
ContainerQueuingLimit queuingLimit =
|
||||
response.getContainerQueuingLimit();
|
||||
if (queuingLimit != null) {
|
||||
context.getContainerManager().updateQueuingLimit(queuingLimit);
|
||||
}
|
||||
}
|
||||
// Handling node resource update case.
|
||||
Resource newResource = response.getResource();
|
||||
if (newResource != null) {
|
||||
updateNMResource(newResource);
|
||||
if (LOG.isDebugEnabled()) {
|
||||
LOG.debug("Node's resource is updated to " +
|
||||
newResource.toString());
|
||||
}
|
||||
}
|
||||
if (YarnConfiguration.timelineServiceV2Enabled(context.getConf())) {
|
||||
updateTimelineClientsAddress(response);
|
||||
}
|
||||
|
||||
} catch (ConnectException e) {
|
||||
//catch and throw the exception if tried MAX wait time to connect RM
|
||||
dispatcher.getEventHandler().handle(
|
||||
new NodeManagerEvent(NodeManagerEventType.SHUTDOWN));
|
||||
// failed to connect to RM.
|
||||
failedToConnect = true;
|
||||
throw new YarnRuntimeException(e);
|
||||
} catch (Exception e) {
|
||||
|
||||
// TODO Better error handling. Thread can die with the rest of the
|
||||
// NM still running.
|
||||
LOG.error("Caught exception in status-updater", e);
|
||||
} finally {
|
||||
synchronized (heartbeatMonitor) {
|
||||
nextHeartBeatInterval = nextHeartBeatInterval <= 0 ?
|
||||
YarnConfiguration.DEFAULT_RM_NM_HEARTBEAT_INTERVAL_MS :
|
||||
nextHeartBeatInterval;
|
||||
try {
|
||||
heartbeatMonitor.wait(nextHeartBeatInterval);
|
||||
} catch (InterruptedException e) {
|
||||
// Do Nothing
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
private void updateTimelineClientsAddress(
|
||||
NodeHeartbeatResponse response) {
|
||||
Map<ApplicationId, String> knownCollectorsMap =
|
||||
response.getAppCollectorsMap();
|
||||
if (knownCollectorsMap == null) {
|
||||
if (LOG.isDebugEnabled()) {
|
||||
LOG.debug("No collectors to update RM");
|
||||
}
|
||||
} else {
|
||||
Set<Map.Entry<ApplicationId, String>> rmKnownCollectors =
|
||||
knownCollectorsMap.entrySet();
|
||||
for (Map.Entry<ApplicationId, String> entry : rmKnownCollectors) {
|
||||
ApplicationId appId = entry.getKey();
|
||||
String collectorAddr = entry.getValue();
|
||||
|
||||
// Only handle applications running on local node.
|
||||
// Not include apps with timeline collectors running in local
|
||||
Application application = context.getApplications().get(appId);
|
||||
// TODO this logic could be problematic if the collector address
|
||||
// gets updated due to NM restart or collector service failure
|
||||
if (application != null &&
|
||||
!context.getRegisteredCollectors().containsKey(appId)) {
|
||||
if (LOG.isDebugEnabled()) {
|
||||
LOG.debug("Sync a new collector address: " + collectorAddr +
|
||||
" for application: " + appId + " from RM.");
|
||||
}
|
||||
NMTimelinePublisher nmTimelinePublisher =
|
||||
context.getNMTimelinePublisher();
|
||||
if (nmTimelinePublisher != null) {
|
||||
nmTimelinePublisher.setTimelineServiceAddress(
|
||||
application.getAppId(), collectorAddr);
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
private void updateMasterKeys(NodeHeartbeatResponse response) {
|
||||
// See if the master-key has rolled over
|
||||
MasterKey updatedMasterKey = response.getContainerTokenMasterKey();
|
||||
if (updatedMasterKey != null) {
|
||||
// Will be non-null only on roll-over on RM side
|
||||
context.getContainerTokenSecretManager().setMasterKey(updatedMasterKey);
|
||||
}
|
||||
|
||||
updatedMasterKey = response.getNMTokenMasterKey();
|
||||
if (updatedMasterKey != null) {
|
||||
context.getNMTokenSecretManager().setMasterKey(updatedMasterKey);
|
||||
}
|
||||
}
|
||||
}
|
||||
}
|
||||
|
||||
@ -213,12 +213,7 @@ public Integer call() {
|
||||
DataOutputStream tokensOutStream = null;
|
||||
|
||||
// Select the working directory for the container
|
||||
Path containerWorkDir =
|
||||
dirsHandler.getLocalPathForWrite(ContainerLocalizer.USERCACHE
|
||||
+ Path.SEPARATOR + user + Path.SEPARATOR
|
||||
+ ContainerLocalizer.APPCACHE + Path.SEPARATOR + appIdStr
|
||||
+ Path.SEPARATOR + containerIdStr,
|
||||
LocalDirAllocator.SIZE_UNKNOWN, false);
|
||||
Path containerWorkDir = deriveContainerWorkDir();
|
||||
recordContainerWorkDir(containerID, containerWorkDir.toString());
|
||||
|
||||
String pidFileSubpath = getPidFileSubpath(appIdStr, containerIdStr);
|
||||
@ -259,12 +254,8 @@ public Integer call() {
|
||||
sanitizeEnv(environment, containerWorkDir, appDirs, userLocalDirs,
|
||||
containerLogDirs, localResources, nmPrivateClasspathJarDir);
|
||||
|
||||
exec.prepareContainer(new ContainerPrepareContext.Builder()
|
||||
.setContainer(container)
|
||||
.setLocalizedResources(localResources)
|
||||
.setUser(user)
|
||||
.setContainerLocalDirs(containerLocalDirs)
|
||||
.setCommands(launchContext.getCommands()).build());
|
||||
prepareContainer(localResources, containerLocalDirs);
|
||||
|
||||
// Write out the environment
|
||||
exec.writeLaunchEnv(containerScriptOutStream, environment,
|
||||
localResources, launchContext.getCommands(),
|
||||
@ -317,6 +308,39 @@ public Integer call() {
|
||||
return ret;
|
||||
}
|
||||
|
||||
private Path deriveContainerWorkDir() throws IOException {
|
||||
|
||||
final String containerWorkDirPath =
|
||||
ContainerLocalizer.USERCACHE +
|
||||
Path.SEPARATOR +
|
||||
container.getUser() +
|
||||
Path.SEPARATOR +
|
||||
ContainerLocalizer.APPCACHE +
|
||||
Path.SEPARATOR +
|
||||
app.getAppId().toString() +
|
||||
Path.SEPARATOR +
|
||||
container.getContainerId().toString();
|
||||
|
||||
final Path containerWorkDir =
|
||||
dirsHandler.getLocalPathForWrite(
|
||||
containerWorkDirPath,
|
||||
LocalDirAllocator.SIZE_UNKNOWN, false);
|
||||
|
||||
return containerWorkDir;
|
||||
}
|
||||
|
||||
private void prepareContainer(Map<Path, List<String>> localResources,
|
||||
List<String> containerLocalDirs) throws IOException {
|
||||
|
||||
exec.prepareContainer(new ContainerPrepareContext.Builder()
|
||||
.setContainer(container)
|
||||
.setLocalizedResources(localResources)
|
||||
.setUser(container.getUser())
|
||||
.setContainerLocalDirs(containerLocalDirs)
|
||||
.setCommands(container.getLaunchContext().getCommands())
|
||||
.build());
|
||||
}
|
||||
|
||||
@SuppressWarnings("unchecked")
|
||||
protected boolean validateContainerState() {
|
||||
// CONTAINER_KILLED_ON_REQUEST should not be missed if the container
|
||||
@ -1117,6 +1141,21 @@ public void sanitizeEnv(Map<String, String> environment, Path pwd,
|
||||
// additional testing. See YARN-358.
|
||||
if (Shell.WINDOWS) {
|
||||
|
||||
sanitizeWindowsEnv(environment, pwd,
|
||||
resources, nmPrivateClasspathJarDir);
|
||||
}
|
||||
// put AuxiliaryService data to environment
|
||||
for (Map.Entry<String, ByteBuffer> meta : containerManager
|
||||
.getAuxServiceMetaData().entrySet()) {
|
||||
AuxiliaryServiceHelper.setServiceDataIntoEnv(
|
||||
meta.getKey(), meta.getValue(), environment);
|
||||
}
|
||||
}
|
||||
|
||||
private void sanitizeWindowsEnv(Map<String, String> environment, Path pwd,
|
||||
Map<Path, List<String>> resources, Path nmPrivateClasspathJarDir)
|
||||
throws IOException {
|
||||
|
||||
String inputClassPath = environment.get(Environment.CLASSPATH.name());
|
||||
|
||||
if (inputClassPath != null && !inputClassPath.isEmpty()) {
|
||||
@ -1148,7 +1187,7 @@ public void sanitizeEnv(Map<String, String> environment, Path pwd,
|
||||
// wildcards to separate classpath entries for each file in the manifest.
|
||||
// To resolve this, append classpath entries explicitly for each
|
||||
// resource.
|
||||
for (Map.Entry<Path,List<String>> entry : resources.entrySet()) {
|
||||
for (Map.Entry<Path, List<String>> entry : resources.entrySet()) {
|
||||
boolean targetIsDirectory = new File(entry.getKey().toUri().getPath())
|
||||
.isDirectory();
|
||||
|
||||
@ -1209,13 +1248,6 @@ public void sanitizeEnv(Map<String, String> environment, Path pwd,
|
||||
environment.put(Environment.CLASSPATH.name(), replacementClassPath);
|
||||
}
|
||||
}
|
||||
// put AuxiliaryService data to environment
|
||||
for (Map.Entry<String, ByteBuffer> meta : containerManager
|
||||
.getAuxServiceMetaData().entrySet()) {
|
||||
AuxiliaryServiceHelper.setServiceDataIntoEnv(
|
||||
meta.getKey(), meta.getValue(), environment);
|
||||
}
|
||||
}
|
||||
|
||||
public static String getExitCodeFile(String pidFile) {
|
||||
return pidFile + EXIT_CODE_FILE_SUFFIX;
|
||||
|
||||
@ -424,10 +424,6 @@ public void launchContainer(ContainerRuntimeContext ctx)
|
||||
//List<String> -> stored as List -> fetched/converted to List<String>
|
||||
//we can't do better here thanks to type-erasure
|
||||
@SuppressWarnings("unchecked")
|
||||
List<String> localDirs = ctx.getExecutionAttribute(LOCAL_DIRS);
|
||||
@SuppressWarnings("unchecked")
|
||||
List<String> logDirs = ctx.getExecutionAttribute(LOG_DIRS);
|
||||
@SuppressWarnings("unchecked")
|
||||
List<String> filecacheDirs = ctx.getExecutionAttribute(FILECACHE_DIRS);
|
||||
@SuppressWarnings("unchecked")
|
||||
List<String> containerLocalDirs = ctx.getExecutionAttribute(
|
||||
@ -489,9 +485,6 @@ public void launchContainer(ContainerRuntimeContext ctx)
|
||||
|
||||
addCGroupParentIfRequired(resourcesOpts, containerIdStr, runCommand);
|
||||
|
||||
Path nmPrivateContainerScriptPath = ctx.getExecutionAttribute(
|
||||
NM_PRIVATE_CONTAINER_SCRIPT_PATH);
|
||||
|
||||
String disableOverride = environment.get(
|
||||
ENV_DOCKER_CONTAINER_RUN_OVERRIDE_DISABLE);
|
||||
|
||||
@ -511,33 +504,8 @@ public void launchContainer(ContainerRuntimeContext ctx)
|
||||
|
||||
String commandFile = dockerClient.writeCommandToTempFile(runCommand,
|
||||
containerIdStr);
|
||||
PrivilegedOperation launchOp = new PrivilegedOperation(
|
||||
PrivilegedOperation.OperationType.LAUNCH_DOCKER_CONTAINER);
|
||||
|
||||
launchOp.appendArgs(runAsUser, ctx.getExecutionAttribute(USER),
|
||||
Integer.toString(PrivilegedOperation
|
||||
.RunAsUserCommand.LAUNCH_DOCKER_CONTAINER.getValue()),
|
||||
ctx.getExecutionAttribute(APPID),
|
||||
containerIdStr, containerWorkDir.toString(),
|
||||
nmPrivateContainerScriptPath.toUri().getPath(),
|
||||
ctx.getExecutionAttribute(NM_PRIVATE_TOKENS_PATH).toUri().getPath(),
|
||||
ctx.getExecutionAttribute(PID_FILE_PATH).toString(),
|
||||
StringUtils.join(PrivilegedOperation.LINUX_FILE_PATH_SEPARATOR,
|
||||
localDirs),
|
||||
StringUtils.join(PrivilegedOperation.LINUX_FILE_PATH_SEPARATOR,
|
||||
logDirs),
|
||||
commandFile,
|
||||
resourcesOpts);
|
||||
|
||||
String tcCommandFile = ctx.getExecutionAttribute(TC_COMMAND_FILE);
|
||||
|
||||
if (tcCommandFile != null) {
|
||||
launchOp.appendArgs(tcCommandFile);
|
||||
}
|
||||
if (LOG.isDebugEnabled()) {
|
||||
LOG.debug("Launching container with cmd: " + runCommand
|
||||
.getCommandWithArguments());
|
||||
}
|
||||
PrivilegedOperation launchOp = buildLaunchOp(ctx,
|
||||
commandFile, runCommand);
|
||||
|
||||
try {
|
||||
privilegedOperationExecutor.executePrivilegedOperation(null,
|
||||
@ -635,4 +603,53 @@ public String[] getIpAndHost(Container container) {
|
||||
}
|
||||
return null;
|
||||
}
|
||||
|
||||
|
||||
|
||||
private PrivilegedOperation buildLaunchOp(ContainerRuntimeContext ctx,
|
||||
String commandFile, DockerRunCommand runCommand) {
|
||||
|
||||
String runAsUser = ctx.getExecutionAttribute(RUN_AS_USER);
|
||||
String containerIdStr = ctx.getContainer().getContainerId().toString();
|
||||
Path nmPrivateContainerScriptPath = ctx.getExecutionAttribute(
|
||||
NM_PRIVATE_CONTAINER_SCRIPT_PATH);
|
||||
Path containerWorkDir = ctx.getExecutionAttribute(CONTAINER_WORK_DIR);
|
||||
//we can't do better here thanks to type-erasure
|
||||
@SuppressWarnings("unchecked")
|
||||
List<String> localDirs = ctx.getExecutionAttribute(LOCAL_DIRS);
|
||||
@SuppressWarnings("unchecked")
|
||||
List<String> logDirs = ctx.getExecutionAttribute(LOG_DIRS);
|
||||
String resourcesOpts = ctx.getExecutionAttribute(RESOURCES_OPTIONS);
|
||||
|
||||
PrivilegedOperation launchOp = new PrivilegedOperation(
|
||||
PrivilegedOperation.OperationType.LAUNCH_DOCKER_CONTAINER);
|
||||
|
||||
launchOp.appendArgs(runAsUser, ctx.getExecutionAttribute(USER),
|
||||
Integer.toString(PrivilegedOperation
|
||||
.RunAsUserCommand.LAUNCH_DOCKER_CONTAINER.getValue()),
|
||||
ctx.getExecutionAttribute(APPID),
|
||||
containerIdStr,
|
||||
containerWorkDir.toString(),
|
||||
nmPrivateContainerScriptPath.toUri().getPath(),
|
||||
ctx.getExecutionAttribute(NM_PRIVATE_TOKENS_PATH).toUri().getPath(),
|
||||
ctx.getExecutionAttribute(PID_FILE_PATH).toString(),
|
||||
StringUtils.join(PrivilegedOperation.LINUX_FILE_PATH_SEPARATOR,
|
||||
localDirs),
|
||||
StringUtils.join(PrivilegedOperation.LINUX_FILE_PATH_SEPARATOR,
|
||||
logDirs),
|
||||
commandFile,
|
||||
resourcesOpts);
|
||||
|
||||
String tcCommandFile = ctx.getExecutionAttribute(TC_COMMAND_FILE);
|
||||
|
||||
if (tcCommandFile != null) {
|
||||
launchOp.appendArgs(tcCommandFile);
|
||||
}
|
||||
if (LOG.isDebugEnabled()) {
|
||||
LOG.debug("Launching container with cmd: " + runCommand
|
||||
.getCommandWithArguments());
|
||||
}
|
||||
|
||||
return launchOp;
|
||||
}
|
||||
}
|
||||
|
||||
Loading…
Reference in New Issue
Block a user