Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view

Large diffs are not rendered by default.

Original file line number Diff line number Diff line change
Expand Up @@ -21,11 +21,13 @@

import org.apache.iotdb.commons.consensus.index.impl.MinimumProgressIndex;
import org.apache.iotdb.commons.exception.IllegalPathException;
import org.apache.iotdb.commons.pipe.agent.plugin.builtin.BuiltinPipePlugin;
import org.apache.iotdb.commons.pipe.agent.task.PipeTaskAgent;
import org.apache.iotdb.commons.pipe.agent.task.meta.PipeStaticMeta;
import org.apache.iotdb.commons.pipe.agent.task.meta.PipeTaskMeta;
import org.apache.iotdb.commons.pipe.agent.task.meta.PipeType;
import org.apache.iotdb.commons.pipe.config.constant.PipeSinkConstant;
import org.apache.iotdb.commons.pipe.config.constant.PipeSourceConstant;
import org.apache.iotdb.commons.pipe.config.constant.SystemConstant;
import org.apache.iotdb.db.pipe.agent.task.PipeDataNodeTask;
import org.apache.iotdb.db.pipe.agent.task.execution.PipeProcessorSubtaskExecutor;
Expand Down Expand Up @@ -64,25 +66,22 @@ public class PipeDataNodeTaskBuilder {
private static final PipeProcessorSubtaskExecutor PROCESSOR_EXECUTOR =
PipeSubtaskExecutorManager.getInstance().getProcessorExecutor();

protected final Map<String, String> systemParameters = new HashMap<>();

public PipeDataNodeTaskBuilder(
final PipeStaticMeta pipeStaticMeta, final int regionId, final PipeTaskMeta pipeTaskMeta) {
this.pipeStaticMeta = pipeStaticMeta;
this.regionId = regionId;
this.pipeTaskMeta = pipeTaskMeta;
generateSystemParameters();
}

public PipeDataNodeTask build() {
// Event flow: source -> processor -> sink

// Analyzes the PipeParameters to identify potential conflicts.
final PipeParameters sourceParameters =
blendUserAndSystemParameters(pipeStaticMeta.getExtractorParameters());
blendUserAndSystemParameters(pipeStaticMeta.getExtractorParameters(), pipeTaskMeta);
final PipeParameters sinkParameters =
blendUserAndSystemParameters(pipeStaticMeta.getConnectorParameters());
checkConflict(sourceParameters, sinkParameters);
blendUserAndSystemParameters(pipeStaticMeta.getConnectorParameters(), pipeTaskMeta);
preprocessParameters(sourceParameters, sinkParameters);

// We first build the source and sink, then build the processor.
final PipeTaskSourceStage sourceStage =
Expand Down Expand Up @@ -121,7 +120,7 @@ public PipeDataNodeTask build() {
new PipeTaskProcessorStage(
pipeStaticMeta.getPipeName(),
pipeStaticMeta.getCreationTime(),
blendUserAndSystemParameters(pipeStaticMeta.getProcessorParameters()),
blendUserAndSystemParameters(pipeStaticMeta.getProcessorParameters(), pipeTaskMeta),
regionId,
sourceStage.getEventSupplier(),
sinkStage.getPipeSinkPendingQueue(),
Expand All @@ -139,22 +138,25 @@ public PipeDataNodeTask build() {
pipeStaticMeta.getPipeName(), regionId, sourceStage, processorStage, sinkStage);
}

private void generateSystemParameters() {
public static PipeParameters blendUserAndSystemParameters(
final PipeParameters userParameters, final PipeTaskMeta pipeTaskMeta) {
// Deep copy the user parameters to avoid modification of the original parameters.
// If the original parameters are modified, progress index report will be affected.
final Map<String, String> blendedParameters = new HashMap<>(userParameters.getAttribute());
if (!(pipeTaskMeta.getProgressIndex() instanceof MinimumProgressIndex)
|| pipeTaskMeta.isNewlyAdded()) {
systemParameters.put(SystemConstant.RESTART_OR_NEWLY_ADDED_KEY, Boolean.TRUE.toString());
blendedParameters.put(SystemConstant.RESTART_OR_NEWLY_ADDED_KEY, Boolean.TRUE.toString());
}
return new PipeParameters(blendedParameters);
}

private PipeParameters blendUserAndSystemParameters(final PipeParameters userParameters) {
// Deep copy the user parameters to avoid modification of the original parameters.
// If the original parameters are modified, progress index report will be affected.
final Map<String, String> blendedParameters = new HashMap<>(userParameters.getAttribute());
blendedParameters.putAll(systemParameters);
return new PipeParameters(blendedParameters);
public static void preprocessParameters(
final PipeParameters sourceParameters, final PipeParameters sinkParameters) {
checkConflict(sourceParameters, sinkParameters);
injectParameters(sourceParameters, sinkParameters);
}

private void checkConflict(
private static void checkConflict(
final PipeParameters sourceParameters, final PipeParameters sinkParameters) {
final Pair<Boolean, Boolean> insertionDeletionListeningOptionPair;
final boolean shouldTerminatePipeOnAllHistoricalEventsConsumed;
Expand Down Expand Up @@ -220,4 +222,30 @@ private void checkConflict(
}
}
}

private static void injectParameters(
final PipeParameters sourceParameters, final PipeParameters sinkParameters) {
final boolean isSourceExternal =
!BuiltinPipePlugin.BUILTIN_SOURCES.contains(
sourceParameters
.getStringOrDefault(
Arrays.asList(PipeSourceConstant.EXTRACTOR_KEY, PipeSourceConstant.SOURCE_KEY),
BuiltinPipePlugin.IOTDB_EXTRACTOR.getPipePluginName())
.toLowerCase());

final String sinkPluginName =
sinkParameters
.getStringOrDefault(
Arrays.asList(PipeSinkConstant.CONNECTOR_KEY, PipeSinkConstant.SINK_KEY),
BuiltinPipePlugin.IOTDB_THRIFT_SINK.getPipePluginName())
.toLowerCase();
final boolean isWriteBackSink =
BuiltinPipePlugin.WRITE_BACK_CONNECTOR.getPipePluginName().equals(sinkPluginName)
|| BuiltinPipePlugin.WRITE_BACK_SINK.getPipePluginName().equals(sinkPluginName);

if (isSourceExternal && isWriteBackSink) {
sinkParameters.addAttribute(
PipeSinkConstant.CONNECTOR_USE_EVENT_USER_NAME_KEY, Boolean.TRUE.toString());
}
}
}
Original file line number Diff line number Diff line change
Expand Up @@ -21,6 +21,7 @@

import org.apache.iotdb.commons.consensus.DataRegionId;
import org.apache.iotdb.commons.pipe.agent.task.connection.UnboundedBlockingPendingQueue;
import org.apache.iotdb.commons.pipe.agent.task.meta.PipeRuntimeMeta;
import org.apache.iotdb.commons.pipe.agent.task.progress.CommitterKey;
import org.apache.iotdb.commons.pipe.agent.task.progress.PipeEventCommitManager;
import org.apache.iotdb.commons.pipe.config.constant.PipeSinkConstant;
Expand Down Expand Up @@ -64,56 +65,26 @@ public synchronized String register(
final Supplier<? extends PipeSinkSubtaskExecutor> executorSupplier,
final PipeParameters pipeSinkParameters,
final PipeTaskSinkRuntimeEnvironment environment) {
final String connectorName =
PipeSinkConstant.getConnectorOrSinkNameWithDefault(pipeSinkParameters);
final String connectorKey =
connectorName
// Convert the value of `CONNECTOR_KEY` or `SINK_KEY` to lowercase
// for matching in `CONNECTOR_CONSTRUCTORS`
.toLowerCase();
final String connectorKey = getConnectorKey(pipeSinkParameters);
PipeEventCommitManager.getInstance()
.register(
environment.getPipeName(),
environment.getCreationTime(),
environment.getRegionId(),
connectorKey);

final boolean isDataSinkConnector =
StorageEngine.getInstance()
.getAllDataRegionIds()
.contains(new DataRegionId(environment.getRegionId()));

final int sinkNum;
final boolean isDataRegionSink = isDataRegionSink(environment.getRegionId());
final int sinkNum = calculateSinkSubtaskNum(pipeSinkParameters, isDataRegionSink, connectorKey);
boolean realTimeFirst = false;
boolean serializeByRegion = false;
String attributeSortedString = generateAttributeSortedString(pipeSinkParameters);
if (isDataSinkConnector) {
serializeByRegion = PipeSinkConstant.isSerializeByRegionEnabled(pipeSinkParameters);
sinkNum =
serializeByRegion
? 1
: pipeSinkParameters.getIntOrDefault(
Arrays.asList(
PipeSinkConstant.CONNECTOR_IOTDB_PARALLEL_TASKS_KEY,
PipeSinkConstant.SINK_IOTDB_PARALLEL_TASKS_KEY),
PipeSinkConstant.SINGLE_THREAD_DEFAULT_SINK.contains(connectorKey)
? 1
: PipeSinkConstant.CONNECTOR_IOTDB_PARALLEL_TASKS_DEFAULT_VALUE);
final String attributeSortedString =
generateAttributeSortedString(pipeSinkParameters, environment.getRegionId());
if (isDataRegionSink) {
realTimeFirst =
pipeSinkParameters.getBooleanOrDefault(
Arrays.asList(
PipeSinkConstant.CONNECTOR_REALTIME_FIRST_KEY,
PipeSinkConstant.SINK_REALTIME_FIRST_KEY),
PipeSinkConstant.CONNECTOR_REALTIME_FIRST_DEFAULT_VALUE);
attributeSortedString =
serializeByRegion
? "data_region_" + environment.getRegionId() + "_" + attributeSortedString
: "data_" + attributeSortedString;
} else {
// Do not allow parallel tasks for schema region connectors
// to avoid the potential disorder of the schema region data transfer
sinkNum = 1;
attributeSortedString = "schema_" + attributeSortedString;
}
environment.setAttributeSortedString(attributeSortedString);

Expand All @@ -135,7 +106,7 @@ public synchronized String register(

for (int connectorIndex = 0; connectorIndex < sinkNum; connectorIndex++) {
final PipeConnector pipeConnector =
isDataSinkConnector
isDataRegionSink
? PipeDataNodeAgent.plugin().dataRegion().reflectSink(pipeSinkParameters)
: PipeDataNodeAgent.plugin().schemaRegion().reflectSink(pipeSinkParameters);
// 1. Construct, validate and customize PipeConnector, and then handshake (create
Expand Down Expand Up @@ -262,7 +233,61 @@ public UnboundedBlockingPendingQueue<Event> getPipeSinkPendingQueue(
.getPendingQueue();
}

private String generateAttributeSortedString(final PipeParameters pipeConnectorParameters) {
public synchronized boolean hasRegisteredSubtasks(
final PipeParameters pipeSinkParameters, final int regionId) {
return attributeSortedString2SubtaskLifeCycleMap.containsKey(
generateAttributeSortedString(pipeSinkParameters, regionId));
}

public static int calculateSinkSubtaskNum(
final PipeParameters pipeSinkParameters, final int regionId) {
final String connectorKey = getConnectorKey(pipeSinkParameters);
return calculateSinkSubtaskNum(pipeSinkParameters, isDataRegionSink(regionId), connectorKey);
}

public static String generateAttributeSortedString(
final PipeParameters pipeSinkParameters, final int regionId) {
final String attributeSortedString = generateAttributeSortedString(pipeSinkParameters);
if (isDataRegionSink(regionId)) {
return PipeSinkConstant.isSerializeByRegionEnabled(pipeSinkParameters)
? "data_region_" + regionId + "_" + attributeSortedString
: "data_" + attributeSortedString;
}
return "schema_" + attributeSortedString;
}

private static String getConnectorKey(final PipeParameters pipeSinkParameters) {
return PipeSinkConstant.getConnectorOrSinkNameWithDefault(pipeSinkParameters).toLowerCase();
}

private static boolean isDataRegionSink(final int regionId) {
return StorageEngine.getInstance().getAllDataRegionIds().contains(new DataRegionId(regionId))
|| PipeRuntimeMeta.isSourceExternal(regionId);
}

private static int calculateSinkSubtaskNum(
final PipeParameters pipeSinkParameters,
final boolean isDataRegionSink,
final String connectorKey) {
if (!isDataRegionSink) {
// Do not allow parallel tasks for schema region connectors to avoid the potential disorder of
// the schema region data transfer.
return 1;
}
if (PipeSinkConstant.isSerializeByRegionEnabled(pipeSinkParameters)) {
return 1;
}
return pipeSinkParameters.getIntOrDefault(
Arrays.asList(
PipeSinkConstant.CONNECTOR_IOTDB_PARALLEL_TASKS_KEY,
PipeSinkConstant.SINK_IOTDB_PARALLEL_TASKS_KEY),
PipeSinkConstant.SINGLE_THREAD_DEFAULT_SINK.contains(connectorKey)
? 1
: PipeSinkConstant.CONNECTOR_IOTDB_PARALLEL_TASKS_DEFAULT_VALUE);
}

private static String generateAttributeSortedString(
final PipeParameters pipeConnectorParameters) {
final TreeMap<String, String> sortedStringSourceMap =
new TreeMap<>(pipeConnectorParameters.getAttribute());
sortedStringSourceMap.remove(SystemConstant.RESTART_OR_NEWLY_ADDED_KEY);
Expand Down
Loading
Loading