@Override
public void onAllNodesAcked(@Nullable Exception e) {
try {
listener.onAllNodesAcked(e);
} catch (Exception inner) {
inner.addSuppressed(e);
logger.error("exception thrown by listener while notifying on all nodes acked", inner);
}
}
@Override
public void onNodeAck(DiscoveryNode node, @Nullable Exception e) {
if (!ackedTaskListener.mustAck(node)) {
// we always wait for the master ack anyway
if (!node.equals(nodes.getMasterNode())) {
return;
}
}
if (e == null) {
logger.trace(
"ack received from node [{}], cluster_state update (version: {})",
node,
clusterStateVersion);
} else {
this.lastFailure = e;
logger.debug(
(Supplier<?>)
() ->
new ParameterizedMessage(
"ack received from node [{}], cluster_state update (version: {})",
node,
clusterStateVersion),
e);
}
if (countDown.countDown()) {
logger.trace(
"all expected nodes acknowledged cluster_state update (version: {})",
clusterStateVersion);
FutureUtils.cancel(ackTimeoutCallback);
ackedTaskListener.onAllNodesAcked(lastFailure);
}
}
AckCountDownListener(
AckedClusterStateTaskListener ackedTaskListener,
long clusterStateVersion,
DiscoveryNodes nodes,
ThreadPool threadPool) {
this.ackedTaskListener = ackedTaskListener;
this.clusterStateVersion = clusterStateVersion;
this.nodes = nodes;
int countDown = 0;
for (DiscoveryNode node : nodes) {
if (ackedTaskListener.mustAck(node)) {
countDown++;
}
}
// we always wait for at least 1 node (the master)
countDown = Math.max(1, countDown);
logger.trace(
"expecting {} acknowledgements for cluster_state update (version: {})",
countDown,
clusterStateVersion);
this.countDown = new CountDown(countDown);
this.ackTimeoutCallback =
threadPool.schedule(
ackedTaskListener.ackTimeout(),
ThreadPool.Names.GENERIC,
new Runnable() {
@Override
public void run() {
onTimeout();
}
});
}
@Override
public void onAckTimeout() {
try {
listener.onAckTimeout();
} catch (Exception e) {
logger.error("exception thrown by listener while notifying on ack timeout", e);
}
}
@Override
public void onTimeout() {
if (countDown.fastForward()) {
logger.trace(
"timeout waiting for acknowledgement for cluster_state update (version: {})",
clusterStateVersion);
ackedTaskListener.onAckTimeout();
}
}
@Override
public TimeValue ackTimeout() {
return listener.ackTimeout();
}
@Override
public boolean mustAck(DiscoveryNode discoveryNode) {
return listener.mustAck(discoveryNode);
}
<T> void runTasksForExecutor(ClusterStateTaskExecutor<T> executor) {
final ArrayList<UpdateTask<T>> toExecute = new ArrayList<>();
final Map<String, ArrayList<T>> processTasksBySource = new HashMap<>();
synchronized (updateTasksPerExecutor) {
List<UpdateTask> pending = updateTasksPerExecutor.remove(executor);
if (pending != null) {
for (UpdateTask<T> task : pending) {
if (task.processed.getAndSet(true) == false) {
logger.trace("will process {}", task.toString(executor));
toExecute.add(task);
processTasksBySource
.computeIfAbsent(task.source, s -> new ArrayList<>())
.add(task.task);
} else {
logger.trace("skipping {}, already processed", task.toString(executor));
}
}
}
}
if (toExecute.isEmpty()) {
return;
}
final String tasksSummary =
processTasksBySource
.entrySet()
.stream()
.map(
entry -> {
String tasks = executor.describeTasks(entry.getValue());
return tasks.isEmpty() ? entry.getKey() : entry.getKey() + "[" + tasks + "]";
})
.reduce((s1, s2) -> s1 + ", " + s2)
.orElse("");
if (!lifecycle.started()) {
logger.debug("processing [{}]: ignoring, cluster_service not started", tasksSummary);
return;
}
logger.debug("processing [{}]: execute", tasksSummary);
ClusterState previousClusterState = clusterState;
if (!previousClusterState.nodes().isLocalNodeElectedMaster() && executor.runOnlyOnMaster()) {
logger.debug("failing [{}]: local node is no longer master", tasksSummary);
toExecute.stream().forEach(task -> task.listener.onNoLongerMaster(task.source));
return;
}
ClusterStateTaskExecutor.BatchResult<T> batchResult;
long startTimeNS = currentTimeInNanos();
try {
List<T> inputs =
toExecute.stream().map(tUpdateTask -> tUpdateTask.task).collect(Collectors.toList());
batchResult = executor.execute(previousClusterState, inputs);
} catch (Exception e) {
TimeValue executionTime =
TimeValue.timeValueMillis(
Math.max(0, TimeValue.nsecToMSec(currentTimeInNanos() - startTimeNS)));
if (logger.isTraceEnabled()) {
logger.trace(
(Supplier<?>)
() ->
new ParameterizedMessage(
"failed to execute cluster state update in [{}], state:\nversion [{}], source [{}]\n{}{}{}",
executionTime,
previousClusterState.version(),
tasksSummary,
previousClusterState.nodes().prettyPrint(),
previousClusterState.routingTable().prettyPrint(),
previousClusterState.getRoutingNodes().prettyPrint()),
e);
}
warnAboutSlowTaskIfNeeded(executionTime, tasksSummary);
batchResult =
ClusterStateTaskExecutor.BatchResult.<T>builder()
.failures(toExecute.stream().map(updateTask -> updateTask.task)::iterator, e)
.build(previousClusterState);
}
assert batchResult.executionResults != null;
assert batchResult.executionResults.size() == toExecute.size()
: String.format(
Locale.ROOT,
"expected [%d] task result%s but was [%d]",
toExecute.size(),
toExecute.size() == 1 ? "" : "s",
batchResult.executionResults.size());
boolean assertsEnabled = false;
assert (assertsEnabled = true);
if (assertsEnabled) {
for (UpdateTask<T> updateTask : toExecute) {
assert batchResult.executionResults.containsKey(updateTask.task)
: "missing task result for " + updateTask.toString(executor);
}
}
ClusterState newClusterState = batchResult.resultingState;
final ArrayList<UpdateTask<T>> proccessedListeners = new ArrayList<>();
// fail all tasks that have failed and extract those that are waiting for results
for (UpdateTask<T> updateTask : toExecute) {
assert batchResult.executionResults.containsKey(updateTask.task)
: "missing " + updateTask.toString(executor);
final ClusterStateTaskExecutor.TaskResult executionResult =
batchResult.executionResults.get(updateTask.task);
executionResult.handle(
() -> proccessedListeners.add(updateTask),
ex -> {
logger.debug(
(Supplier<?>)
() ->
new ParameterizedMessage(
"cluster state update task {} failed", updateTask.toString(executor)),
ex);
updateTask.listener.onFailure(updateTask.source, ex);
});
}
if (previousClusterState == newClusterState) {
for (UpdateTask<T> task : proccessedListeners) {
if (task.listener instanceof AckedClusterStateTaskListener) {
// no need to wait for ack if nothing changed, the update can be counted as acknowledged
((AckedClusterStateTaskListener) task.listener).onAllNodesAcked(null);
}
task.listener.clusterStateProcessed(task.source, previousClusterState, newClusterState);
}
TimeValue executionTime =
TimeValue.timeValueMillis(
Math.max(0, TimeValue.nsecToMSec(currentTimeInNanos() - startTimeNS)));
logger.debug(
"processing [{}]: took [{}] no change in cluster_state", tasksSummary, executionTime);
warnAboutSlowTaskIfNeeded(executionTime, tasksSummary);
return;
}
try {
ArrayList<Discovery.AckListener> ackListeners = new ArrayList<>();
if (newClusterState.nodes().isLocalNodeElectedMaster()) {
// only the master controls the version numbers
Builder builder = ClusterState.builder(newClusterState).incrementVersion();
if (previousClusterState.routingTable() != newClusterState.routingTable()) {
builder.routingTable(
RoutingTable.builder(newClusterState.routingTable())
.version(newClusterState.routingTable().version() + 1)
.build());
}
if (previousClusterState.metaData() != newClusterState.metaData()) {
builder.metaData(
MetaData.builder(newClusterState.metaData())
.version(newClusterState.metaData().version() + 1));
}
newClusterState = builder.build();
for (UpdateTask<T> task : proccessedListeners) {
if (task.listener instanceof AckedClusterStateTaskListener) {
final AckedClusterStateTaskListener ackedListener =
(AckedClusterStateTaskListener) task.listener;
if (ackedListener.ackTimeout() == null || ackedListener.ackTimeout().millis() == 0) {
ackedListener.onAckTimeout();
} else {
try {
ackListeners.add(
new AckCountDownListener(
ackedListener,
newClusterState.version(),
newClusterState.nodes(),
threadPool));
} catch (EsRejectedExecutionException ex) {
if (logger.isDebugEnabled()) {
logger.debug(
"Couldn't schedule timeout thread - node might be shutting down", ex);
}
// timeout straightaway, otherwise we could wait forever as the timeout thread has
// not started
ackedListener.onAckTimeout();
}
}
}
}
}
final Discovery.AckListener ackListener = new DelegetingAckListener(ackListeners);
newClusterState.status(ClusterState.ClusterStateStatus.BEING_APPLIED);
if (logger.isTraceEnabled()) {
logger.trace(
"cluster state updated, source [{}]\n{}", tasksSummary, newClusterState.prettyPrint());
} else if (logger.isDebugEnabled()) {
logger.debug(
"cluster state updated, version [{}], source [{}]",
newClusterState.version(),
tasksSummary);
}
ClusterChangedEvent clusterChangedEvent =
new ClusterChangedEvent(tasksSummary, newClusterState, previousClusterState);
// new cluster state, notify all listeners
final DiscoveryNodes.Delta nodesDelta = clusterChangedEvent.nodesDelta();
if (nodesDelta.hasChanges() && logger.isInfoEnabled()) {
String summary = nodesDelta.shortSummary();
if (summary.length() > 0) {
logger.info("{}, reason: {}", summary, tasksSummary);
}
}
nodeConnectionsService.connectToAddedNodes(clusterChangedEvent);
// if we are the master, publish the new state to all nodes
// we publish here before we send a notification to all the listeners, since if it fails
// we don't want to notify
if (newClusterState.nodes().isLocalNodeElectedMaster()) {
logger.debug("publishing cluster state version [{}]", newClusterState.version());
try {
clusterStatePublisher.accept(clusterChangedEvent, ackListener);
} catch (Discovery.FailedToCommitClusterStateException t) {
final long version = newClusterState.version();
logger.warn(
(Supplier<?>)
() ->
new ParameterizedMessage(
"failing [{}]: failed to commit cluster state version [{}]",
tasksSummary,
version),
t);
proccessedListeners.forEach(task -> task.listener.onFailure(task.source, t));
return;
}
}
// update the current cluster state
clusterState = newClusterState;
logger.debug("set local cluster state to version {}", newClusterState.version());
try {
// nothing to do until we actually recover from the gateway or any other block indicates we
// need to disable persistency
if (clusterChangedEvent.state().blocks().disableStatePersistence() == false
&& clusterChangedEvent.metaDataChanged()) {
final Settings incomingSettings = clusterChangedEvent.state().metaData().settings();
clusterSettings.applySettings(incomingSettings);
}
} catch (Exception ex) {
logger.warn("failed to apply cluster settings", ex);
}
for (ClusterStateListener listener : preAppliedListeners) {
try {
listener.clusterChanged(clusterChangedEvent);
} catch (Exception ex) {
logger.warn("failed to notify ClusterStateListener", ex);
}
}
nodeConnectionsService.disconnectFromRemovedNodes(clusterChangedEvent);
newClusterState.status(ClusterState.ClusterStateStatus.APPLIED);
for (ClusterStateListener listener : postAppliedListeners) {
try {
listener.clusterChanged(clusterChangedEvent);
} catch (Exception ex) {
logger.warn("failed to notify ClusterStateListener", ex);
}
}
// manual ack only from the master at the end of the publish
if (newClusterState.nodes().isLocalNodeElectedMaster()) {
try {
ackListener.onNodeAck(newClusterState.nodes().getLocalNode(), null);
} catch (Exception e) {
final DiscoveryNode localNode = newClusterState.nodes().getLocalNode();
logger.debug(
(Supplier<?>)
() ->
new ParameterizedMessage(
"error while processing ack for master node [{}]", localNode),
e);
}
}
for (UpdateTask<T> task : proccessedListeners) {
task.listener.clusterStateProcessed(task.source, previousClusterState, newClusterState);
}
try {
executor.clusterStatePublished(clusterChangedEvent);
} catch (Exception e) {
logger.error(
(Supplier<?>)
() ->
new ParameterizedMessage(
"exception thrown while notifying executor of new cluster state publication [{}]",
tasksSummary),
e);
}
TimeValue executionTime =
TimeValue.timeValueMillis(
Math.max(0, TimeValue.nsecToMSec(currentTimeInNanos() - startTimeNS)));
logger.debug(
"processing [{}]: took [{}] done applying updated cluster_state (version: {}, uuid: {})",
tasksSummary,
executionTime,
newClusterState.version(),
newClusterState.stateUUID());
warnAboutSlowTaskIfNeeded(executionTime, tasksSummary);
} catch (Exception e) {
TimeValue executionTime =
TimeValue.timeValueMillis(
Math.max(0, TimeValue.nsecToMSec(currentTimeInNanos() - startTimeNS)));
final long version = newClusterState.version();
final String stateUUID = newClusterState.stateUUID();
final String prettyPrint = newClusterState.prettyPrint();
logger.warn(
(Supplier<?>)
() ->
new ParameterizedMessage(
"failed to apply updated cluster state in [{}]:\nversion [{}], uuid [{}], source [{}]\n{}",
executionTime,
version,
stateUUID,
tasksSummary,
prettyPrint),
e);
// TODO: do we want to call updateTask.onFailure here?
}
}