kgyrtkirk commented on code in PR #17038:
URL: https://github.com/apache/druid/pull/17038#discussion_r1770980060
##########
extensions-core/multi-stage-query/src/main/java/org/apache/druid/msq/querykit/WindowOperatorQueryFrameProcessor.java:
##########
@@ -510,4 +322,9 @@ private void ensureMaxRowsInAWindowConstraint(int
numRowsInWindow)
));
}
}
+
+ private boolean needToProcessBatch()
+ {
+ return numRowsInFrameRowsAndCols >= maxRowsMaterialized / 2; // Can this
be improved further?
Review Comment:
why divide by `2` ? that doesn't give any guarantee that it will be inside
bounds
people could set it to half if needed - but I think its easier to document
clear things...
##########
extensions-core/multi-stage-query/src/main/java/org/apache/druid/msq/querykit/WindowOperatorQueryFrameProcessor.java:
##########
@@ -154,150 +130,60 @@ public List<WritableFrameChannel> outputChannels()
@Override
public ReturnOrAwait<Object> runIncrementally(IntSet readableInputs)
{
- /*
- There are 2 scenarios:
-
- *** Scenario 1: Query has atleast one window function with an OVER()
clause without a PARTITION BY ***
-
- In this scenario, we add all the RACs to a single RowsAndColumns to be
processed. We do it via ConcatRowsAndColumns, and run all the operators on the
ConcatRowsAndColumns.
- This is done because we anyway need to run the operators on the entire
set of rows when we have an OVER() clause without a PARTITION BY.
- This scenario corresponds to partitionColumnNames.isEmpty()=true code
flow.
-
- *** Scenario 2: All window functions in the query have OVER() clause with
a PARTITION BY ***
-
- In this scenario, we need to process rows for each PARTITION BY group
together, but we can batch multiple PARTITION BY keys into the same RAC before
passing it to the operators for processing.
- Batching is fine since the operators list would have the required
NaivePartitioningOperatorFactory to segregate each PARTITION BY group during
the processing.
-
- The flow for this scenario can be summarised as following:
- 1. Frame Reading and Cursor Initialization: We start by reading a frame
from the inputChannel and initializing frameCursor to iterate over the rows in
that frame.
- 2. Row Comparison: For each row in the frame, we decide whether it
belongs to the same PARTITION BY group as the previous row.
- This is determined by comparePartitionKeys() method.
- Please refer to the Javadoc of that method for further
details and an example illustration.
- 2.1. If the PARTITION BY columns of current row matches the PARTITION
BY columns of the previous row,
- they belong to the same PARTITION BY group, and gets added to
rowsToProcess.
- If the number of total rows materialized exceed
maxRowsMaterialized, we process the pending batch via
processRowsUpToLastPartition() method.
- 2.2. If they don't match, then we have reached a partition boundary.
- In this case, we update the value for lastPartitionIndex.
- 3. End of Input: If the input channel is finished, any remaining rows in
rowsToProcess are processed.
-
- *Illustration of Row Comparison step*
+ if (inputChannel.canRead()) {
+ final Frame frame = inputChannel.read();
+ convertRowFrameToRowsAndColumns(frame);
- Let's say we have window_function() OVER (PARTITION BY A ORDER BY B) in
our query, and we get 3 frames in the input channel to process.
-
- Frame 1
- A, B
- 1, 2
- 1, 3
- 2, 1 --> PARTITION BY key (column A) changed from 1 to 2.
- 2, 2
-
- Frame 2
- A, B
- 3, 1 --> PARTITION BY key (column A) changed from 2 to 3.
- 3, 2
- 3, 3
- 3, 4
-
- Frame 3
- A, B
- 3, 5
- 3, 6
- 4, 1 --> PARTITION BY key (column A) changed from 3 to 4.
- 4, 2
-
- *Why batching?*
- We batch multiple PARTITION BY keys for processing together to avoid the
overhead of creating different RACs for each PARTITION BY keys, as that would
be unnecessary in scenarios where we have a large number of PARTITION BY keys,
but each key having a single row.
-
- *Future thoughts: https://github.com/apache/druid/issues/16126*
- Current approach with R&C and operators materialize a single R&C for
processing. In case of data with low cardinality a single R&C might be too big
to consume. Same for the case of empty OVER() clause.
- Most of the window operations like SUM(), RANK(), RANGE() etc. can be
made with 2 passes of the data. We might think to reimplement them in the MSQ
way so that we do not have to materialize so much data.
- */
-
- if (partitionColumnNames.isEmpty()) {
- // Scenario 1: Query has atleast one window function with an OVER()
clause without a PARTITION BY.
- if (inputChannel.canRead()) {
- final Frame frame = inputChannel.read();
- convertRowFrameToRowsAndColumns(frame);
- return ReturnOrAwait.runAgain();
- } else if (inputChannel.isFinished()) {
- runAllOpsOnMultipleRac(frameRowsAndCols);
- return ReturnOrAwait.returnObject(Unit.instance());
- } else {
- return ReturnOrAwait.awaitAll(inputChannels().size());
- }
- }
-
- // Scenario 2: All window functions in the query have OVER() clause with a
PARTITION BY
- if (frameCursor == null || frameCursor.isDone()) {
- if (readableInputs.isEmpty()) {
- return ReturnOrAwait.awaitAll(1);
- } else if (inputChannel.canRead()) {
- final Frame frame = inputChannel.read();
- frameCursor = FrameProcessors.makeCursor(frame, frameReader);
- makeRowSupplierFromFrameCursor();
- } else if (inputChannel.isFinished()) {
- // Handle any remaining data.
- lastPartitionIndex = rowsToProcess.size() - 1;
- processRowsUpToLastPartition();
- return ReturnOrAwait.returnObject(Unit.instance());
- } else {
- return ReturnOrAwait.runAgain();
- }
- }
-
- while (!frameCursor.isDone()) {
- final ResultRow currentRow = rowSupplierFromFrameCursor.get();
- if (outputRow == null) {
- outputRow = currentRow;
- rowsToProcess.add(currentRow);
- } else if (comparePartitionKeys(outputRow, currentRow,
partitionColumnNames)) {
- // Add current row to the same batch of rows for processing.
- rowsToProcess.add(currentRow);
- if (rowsToProcess.size() > maxRowsMaterialized) {
- // We don't want to materialize more than maxRowsMaterialized rows
at any point in time, so process the pending batch.
- processRowsUpToLastPartition();
+ if (needToProcessBatch()) {
+ runAllOpsOnBatch();
+ try {
+ flushAllRowsAndCols(resultRowAndCols);
+ }
+ catch (IOException e) {
+ throw new RuntimeException(e);
}
- ensureMaxRowsInAWindowConstraint(rowsToProcess.size());
- } else {
- lastPartitionIndex = rowsToProcess.size() - 1;
- outputRow = currentRow.copy();
- return ReturnOrAwait.runAgain();
}
- frameCursor.advance();
+ return ReturnOrAwait.runAgain();
+ } else if (inputChannel.isFinished()) {
+ runAllOpsOnBatch();
+ return ReturnOrAwait.returnObject(Unit.instance());
+ } else {
+ return ReturnOrAwait.awaitAll(inputChannels().size());
}
- return ReturnOrAwait.runAgain();
}
- /**
- * @param listOfRacs Concat this list of {@link RowsAndColumns} to a {@link
ConcatRowsAndColumns} to use as a single input for the operators to be run
- */
- private void runAllOpsOnMultipleRac(ArrayList<RowsAndColumns> listOfRacs)
+ private void initialiseOperator()
{
- Operator op = new Operator()
+ op = new Operator()
{
@Nullable
@Override
public Closeable goOrContinue(Closeable continuationObject, Receiver
receiver)
{
- RowsAndColumns rac = new ConcatRowsAndColumns(listOfRacs);
+ RowsAndColumns rac = new ConcatRowsAndColumns(new
ArrayList<>(frameRowsAndCols));
+ frameRowsAndCols.clear();
+ numRowsInFrameRowsAndCols = 0;
ensureMaxRowsInAWindowConstraint(rac.numRows());
receiver.push(rac);
- receiver.completed();
- return null;
+
+ if (inputChannel.isFinished()) {
+ // Only call completed() when the input channel is finished.
+ receiver.completed();
+ return null; // Signal that the operator has completed its work
+ }
+
+ // Return a non-null continuation object to indicate that we want to
continue processing.
+ return () -> {};
}
};
- runOperatorsAfterThis(op);
- }
-
- /**
- * @param op Base operator for the operators to be run. Other operators are
wrapped under this to run
- */
- private void runOperatorsAfterThis(Operator op)
- {
for (OperatorFactory of : operatorFactoryList) {
op = of.wrap(op);
}
- Operator.go(op, new Operator.Receiver()
+ }
+
+ private void runAllOpsOnBatch()
+ {
+ op.goOrContinue(null, new Operator.Receiver()
Review Comment:
I find this a bit wierd....
* why the `flush` method is outside of this class?
* why this have to cache all the rows?
* why it have to delay write up-until `completed` ?
##########
extensions-core/multi-stage-query/src/main/java/org/apache/druid/msq/querykit/WindowOperatorQueryFrameProcessorFactory.java:
##########
@@ -61,26 +60,19 @@ public class WindowOperatorQueryFrameProcessorFactory
extends BaseFrameProcessor
private final List<OperatorFactory> operatorList;
private final RowSignature stageRowSignature;
private final int maxRowsMaterializedInWindow;
- private final List<String> partitionColumnNames;
@JsonCreator
public WindowOperatorQueryFrameProcessorFactory(
@JsonProperty("query") WindowOperatorQuery query,
@JsonProperty("operatorList") List<OperatorFactory> operatorFactoryList,
@JsonProperty("stageRowSignature") RowSignature stageRowSignature,
- @JsonProperty("maxRowsMaterializedInWindow") int
maxRowsMaterializedInWindow,
- @JsonProperty("partitionColumnNames") List<String> partitionColumnNames
+ @JsonProperty("maxRowsMaterializedInWindow") int
maxRowsMaterializedInWindow
Review Comment:
does this really needs to be an option to the `processorFactory` ?
isn't this a system-wide or query-wide setting? I don't see any reason to
make it any finer grain than that..
##########
processing/src/main/java/org/apache/druid/query/operator/BasePartitioningOperator.java:
##########
@@ -0,0 +1,121 @@
+/*
+ * Licensed to the Apache Software Foundation (ASF) under one
+ * or more contributor license agreements. See the NOTICE file
+ * distributed with this work for additional information
+ * regarding copyright ownership. The ASF licenses this file
+ * to you under the Apache License, Version 2.0 (the
+ * "License"); you may not use this file except in compliance
+ * with the License. You may obtain a copy of the License at
+ *
+ * http://www.apache.org/licenses/LICENSE-2.0
+ *
+ * Unless required by applicable law or agreed to in writing,
+ * software distributed under the License is distributed on an
+ * "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
+ * KIND, either express or implied. See the License for the
+ * specific language governing permissions and limitations
+ * under the License.
+ */
+
+package org.apache.druid.query.operator;
+
+import org.apache.druid.error.DruidException;
+import org.apache.druid.java.util.common.RE;
+import org.apache.druid.query.rowsandcols.RowsAndColumns;
+
+import java.io.Closeable;
+import java.io.IOException;
+import java.util.Iterator;
+import java.util.List;
+import java.util.concurrent.atomic.AtomicReference;
+
+public abstract class BasePartitioningOperator implements Operator
Review Comment:
can we call it `Abstract*` as its abstract?
##########
extensions-core/multi-stage-query/src/main/java/org/apache/druid/msq/querykit/WindowOperatorQueryFrameProcessor.java:
##########
@@ -338,6 +218,9 @@ private void flushAllRowsAndCols(ArrayList<RowsAndColumns>
resultRowAndCols) thr
AtomicInteger rowId = new AtomicInteger(0);
createFrameWriterIfNeeded(rac, rowId);
writeRacToFrame(rac, rowId);
Review Comment:
seems like there are some incorrect method contracts here: as `rowId` is
used incorrectly:
* if this method would be called 2 times the 1st invocation of
`createFrameWriterIfNeeded` would still retain the previous rowid reference
* however `writeRacToFrame` would use the one passed from here
##########
extensions-core/multi-stage-query/src/main/java/org/apache/druid/msq/querykit/WindowOperatorQueryKit.java:
##########
@@ -377,4 +324,37 @@ private static RowSignature
computeSignatureForFinalWindowStage(RowSignature row
finalWindowClusterBy.getColumns()
);
}
+
+ /**
+ * This method converts the operator chain received from native plan into
MSQ plan.
+ * (NaiveSortOperator -> Naive/GlueingPartitioningOperator ->
WindowOperator) is converted into (GlueingPartitioningOperator ->
PartitionSortOperator -> WindowOperator).
+ * We rely on MSQ's shuffling to do the clustering on partitioning keys for
us at every stage.
+ * This conversion allows us to blindly read N rows from input channel and
push them into the operator chain, and repeat until the input channel isn't
finished.
+ * @param operatorFactoryListFromQuery
+ * @param maxRowsMaterializedInWindow
+ * @return
+ */
+ private List<OperatorFactory>
getOperatorFactoryListForStageDefinition(List<OperatorFactory>
operatorFactoryListFromQuery, int maxRowsMaterializedInWindow)
Review Comment:
please open a preceeding PR which fixes the optimization model of the
WindowOperatorQueryKit to use a struct which is like:
```
class StageX {
SortOperator sortOperator;
PartitionOperator partitionOperator;
List<Operator> operators
}
```
##########
extensions-core/multi-stage-query/src/main/java/org/apache/druid/msq/querykit/WindowOperatorQueryKit.java:
##########
@@ -377,4 +324,37 @@ private static RowSignature
computeSignatureForFinalWindowStage(RowSignature row
finalWindowClusterBy.getColumns()
);
}
+
+ /**
+ * This method converts the operator chain received from native plan into
MSQ plan.
+ * (NaiveSortOperator -> Naive/GlueingPartitioningOperator ->
WindowOperator) is converted into (GlueingPartitioningOperator ->
PartitionSortOperator -> WindowOperator).
+ * We rely on MSQ's shuffling to do the clustering on partitioning keys for
us at every stage.
+ * This conversion allows us to blindly read N rows from input channel and
push them into the operator chain, and repeat until the input channel isn't
finished.
+ * @param operatorFactoryListFromQuery
+ * @param maxRowsMaterializedInWindow
+ * @return
+ */
+ private List<OperatorFactory>
getOperatorFactoryListForStageDefinition(List<OperatorFactory>
operatorFactoryListFromQuery, int maxRowsMaterializedInWindow)
+ {
+ final List<OperatorFactory> operatorFactoryList = new ArrayList<>();
+ final List<OperatorFactory> sortOperatorFactoryList = new ArrayList<>();
+ for (OperatorFactory operatorFactory : operatorFactoryListFromQuery) {
+ if (operatorFactory instanceof BasePartitioningOperatorFactory) {
+ BasePartitioningOperatorFactory partition =
(BasePartitioningOperatorFactory) operatorFactory;
+ operatorFactoryList.add(new
GlueingPartitioningOperatorFactory(partition.getPartitionColumns(),
maxRowsMaterializedInWindow));
+ } else if (operatorFactory instanceof BaseSortOperatorFactory) {
+ BaseSortOperatorFactory sortOperatorFactory =
(BaseSortOperatorFactory) operatorFactory;
+ sortOperatorFactoryList.add(new
PartitionSortOperatorFactory(sortOperatorFactory.getSortColumns()));
Review Comment:
this should be a planner level decision - if such thing is done here you
will eventully be doing some optimization jobs here.
please back off from this in MSQ
##########
extensions-core/multi-stage-query/src/main/java/org/apache/druid/msq/querykit/WindowOperatorQueryFrameProcessor.java:
##########
@@ -154,150 +130,60 @@ public List<WritableFrameChannel> outputChannels()
@Override
public ReturnOrAwait<Object> runIncrementally(IntSet readableInputs)
{
- /*
- There are 2 scenarios:
-
- *** Scenario 1: Query has atleast one window function with an OVER()
clause without a PARTITION BY ***
-
- In this scenario, we add all the RACs to a single RowsAndColumns to be
processed. We do it via ConcatRowsAndColumns, and run all the operators on the
ConcatRowsAndColumns.
- This is done because we anyway need to run the operators on the entire
set of rows when we have an OVER() clause without a PARTITION BY.
- This scenario corresponds to partitionColumnNames.isEmpty()=true code
flow.
-
- *** Scenario 2: All window functions in the query have OVER() clause with
a PARTITION BY ***
-
- In this scenario, we need to process rows for each PARTITION BY group
together, but we can batch multiple PARTITION BY keys into the same RAC before
passing it to the operators for processing.
- Batching is fine since the operators list would have the required
NaivePartitioningOperatorFactory to segregate each PARTITION BY group during
the processing.
-
- The flow for this scenario can be summarised as following:
- 1. Frame Reading and Cursor Initialization: We start by reading a frame
from the inputChannel and initializing frameCursor to iterate over the rows in
that frame.
- 2. Row Comparison: For each row in the frame, we decide whether it
belongs to the same PARTITION BY group as the previous row.
- This is determined by comparePartitionKeys() method.
- Please refer to the Javadoc of that method for further
details and an example illustration.
- 2.1. If the PARTITION BY columns of current row matches the PARTITION
BY columns of the previous row,
- they belong to the same PARTITION BY group, and gets added to
rowsToProcess.
- If the number of total rows materialized exceed
maxRowsMaterialized, we process the pending batch via
processRowsUpToLastPartition() method.
- 2.2. If they don't match, then we have reached a partition boundary.
- In this case, we update the value for lastPartitionIndex.
- 3. End of Input: If the input channel is finished, any remaining rows in
rowsToProcess are processed.
-
- *Illustration of Row Comparison step*
+ if (inputChannel.canRead()) {
+ final Frame frame = inputChannel.read();
+ convertRowFrameToRowsAndColumns(frame);
- Let's say we have window_function() OVER (PARTITION BY A ORDER BY B) in
our query, and we get 3 frames in the input channel to process.
-
- Frame 1
- A, B
- 1, 2
- 1, 3
- 2, 1 --> PARTITION BY key (column A) changed from 1 to 2.
- 2, 2
-
- Frame 2
- A, B
- 3, 1 --> PARTITION BY key (column A) changed from 2 to 3.
- 3, 2
- 3, 3
- 3, 4
-
- Frame 3
- A, B
- 3, 5
- 3, 6
- 4, 1 --> PARTITION BY key (column A) changed from 3 to 4.
- 4, 2
-
- *Why batching?*
- We batch multiple PARTITION BY keys for processing together to avoid the
overhead of creating different RACs for each PARTITION BY keys, as that would
be unnecessary in scenarios where we have a large number of PARTITION BY keys,
but each key having a single row.
-
- *Future thoughts: https://github.com/apache/druid/issues/16126*
- Current approach with R&C and operators materialize a single R&C for
processing. In case of data with low cardinality a single R&C might be too big
to consume. Same for the case of empty OVER() clause.
- Most of the window operations like SUM(), RANK(), RANGE() etc. can be
made with 2 passes of the data. We might think to reimplement them in the MSQ
way so that we do not have to materialize so much data.
- */
-
- if (partitionColumnNames.isEmpty()) {
- // Scenario 1: Query has atleast one window function with an OVER()
clause without a PARTITION BY.
- if (inputChannel.canRead()) {
- final Frame frame = inputChannel.read();
- convertRowFrameToRowsAndColumns(frame);
- return ReturnOrAwait.runAgain();
- } else if (inputChannel.isFinished()) {
- runAllOpsOnMultipleRac(frameRowsAndCols);
- return ReturnOrAwait.returnObject(Unit.instance());
- } else {
- return ReturnOrAwait.awaitAll(inputChannels().size());
- }
- }
-
- // Scenario 2: All window functions in the query have OVER() clause with a
PARTITION BY
- if (frameCursor == null || frameCursor.isDone()) {
- if (readableInputs.isEmpty()) {
- return ReturnOrAwait.awaitAll(1);
- } else if (inputChannel.canRead()) {
- final Frame frame = inputChannel.read();
- frameCursor = FrameProcessors.makeCursor(frame, frameReader);
- makeRowSupplierFromFrameCursor();
- } else if (inputChannel.isFinished()) {
- // Handle any remaining data.
- lastPartitionIndex = rowsToProcess.size() - 1;
- processRowsUpToLastPartition();
- return ReturnOrAwait.returnObject(Unit.instance());
- } else {
- return ReturnOrAwait.runAgain();
- }
- }
-
- while (!frameCursor.isDone()) {
- final ResultRow currentRow = rowSupplierFromFrameCursor.get();
- if (outputRow == null) {
- outputRow = currentRow;
- rowsToProcess.add(currentRow);
- } else if (comparePartitionKeys(outputRow, currentRow,
partitionColumnNames)) {
- // Add current row to the same batch of rows for processing.
- rowsToProcess.add(currentRow);
- if (rowsToProcess.size() > maxRowsMaterialized) {
- // We don't want to materialize more than maxRowsMaterialized rows
at any point in time, so process the pending batch.
- processRowsUpToLastPartition();
+ if (needToProcessBatch()) {
+ runAllOpsOnBatch();
+ try {
+ flushAllRowsAndCols(resultRowAndCols);
+ }
+ catch (IOException e) {
+ throw new RuntimeException(e);
}
- ensureMaxRowsInAWindowConstraint(rowsToProcess.size());
- } else {
- lastPartitionIndex = rowsToProcess.size() - 1;
- outputRow = currentRow.copy();
- return ReturnOrAwait.runAgain();
}
- frameCursor.advance();
+ return ReturnOrAwait.runAgain();
+ } else if (inputChannel.isFinished()) {
+ runAllOpsOnBatch();
+ return ReturnOrAwait.returnObject(Unit.instance());
+ } else {
+ return ReturnOrAwait.awaitAll(inputChannels().size());
}
- return ReturnOrAwait.runAgain();
}
- /**
- * @param listOfRacs Concat this list of {@link RowsAndColumns} to a {@link
ConcatRowsAndColumns} to use as a single input for the operators to be run
- */
- private void runAllOpsOnMultipleRac(ArrayList<RowsAndColumns> listOfRacs)
+ private void initialiseOperator()
{
- Operator op = new Operator()
+ op = new Operator()
{
@Nullable
@Override
public Closeable goOrContinue(Closeable continuationObject, Receiver
receiver)
{
- RowsAndColumns rac = new ConcatRowsAndColumns(listOfRacs);
+ RowsAndColumns rac = new ConcatRowsAndColumns(new
ArrayList<>(frameRowsAndCols));
+ frameRowsAndCols.clear();
+ numRowsInFrameRowsAndCols = 0;
Review Comment:
coupled contract of `numRowsInFrameRowsAndCols` + `frameRowsAndCols` ; make
a small builder class which has an:
* add racs
* can answer numRows
* production assembles the actual RAC
##########
extensions-core/multi-stage-query/src/main/java/org/apache/druid/msq/querykit/WindowOperatorQueryKit.java:
##########
@@ -120,127 +111,83 @@ public QueryDefinition makeQueryDefinition(
maxRowsMaterialized = Limits.MAX_ROWS_MATERIALIZED_IN_WINDOW;
Review Comment:
evaluating such things as defaults should be out-of-scope in this class
move this conditional to somewhere like `MultiStageQueryContext.getMax...`
##########
extensions-core/multi-stage-query/src/main/java/org/apache/druid/msq/querykit/WindowOperatorQueryKit.java:
##########
@@ -120,127 +111,83 @@ public QueryDefinition makeQueryDefinition(
maxRowsMaterialized = Limits.MAX_ROWS_MATERIALIZED_IN_WINDOW;
}
- if (isEmptyOverPresent) {
- // Move everything to a single partition since we have to load all the
data on a single worker anyway to compute empty over() clause.
- log.info(
- "Empty over clause is present in the query. Creating a single stage
with all operator factories [%s].",
- queryToRun.getOperators()
- );
- queryDefBuilder.add(
- StageDefinition.builder(firstStageNumber)
- .inputs(new StageInputSpec(firstStageNumber - 1))
- .signature(finalWindowStageRowSignature)
- .maxWorkerCount(maxWorkerCount)
- .shuffleSpec(finalWindowStageShuffleSpec)
- .processorFactory(new
WindowOperatorQueryFrameProcessorFactory(
- queryToRun,
- queryToRun.getOperators(),
- finalWindowStageRowSignature,
- maxRowsMaterialized,
- Collections.emptyList()
- ))
- );
- } else {
- // There are multiple windows present in the query.
- // Create stages for each window in the query.
- // These stages will be serialized.
- // The partition by clause of the next window will be the shuffle key
for the previous window.
- RowSignature.Builder bob = RowSignature.builder();
- RowSignature signatureFromInput =
dataSourcePlan.getSubQueryDefBuilder().get().build().getFinalStageDefinition().getSignature();
- log.info("Row signature received from last stage is [%s].",
signatureFromInput);
-
- for (int i = 0; i < signatureFromInput.getColumnNames().size(); i++) {
- bob.add(signatureFromInput.getColumnName(i),
signatureFromInput.getColumnType(i).get());
- }
+ // There are multiple windows present in the query.
+ // Create stages for each window in the query.
+ // These stages will be serialized.
+ // The partition by clause of the next window will be the shuffle key for
the previous window.
+ RowSignature.Builder bob = RowSignature.builder();
+ RowSignature signatureFromInput =
dataSourcePlan.getSubQueryDefBuilder().get().build().getFinalStageDefinition().getSignature();
+ log.info("Row signature received from last stage is [%s].",
signatureFromInput);
+
+ for (int i = 0; i < signatureFromInput.getColumnNames().size(); i++) {
+ bob.add(signatureFromInput.getColumnName(i),
signatureFromInput.getColumnType(i).get());
+ }
- List<String> partitionColumnNames = new ArrayList<>();
-
- /*
- operatorList is a List<List<OperatorFactory>>, where each
List<OperatorFactory> corresponds to the operator factories
- to be used for a different window stage.
-
- We iterate over operatorList, and add the definition for a window stage
to QueryDefinitionBuilder.
- */
- for (int i = 0; i < operatorList.size(); i++) {
- for (OperatorFactory operatorFactory : operatorList.get(i)) {
- if (operatorFactory instanceof WindowOperatorFactory) {
- List<String> outputColumnNames = ((WindowOperatorFactory)
operatorFactory).getProcessor().getOutputColumnNames();
-
- // Need to add column names which are present in outputColumnNames
and rowSignature but not in bob,
- // since they need to be present in the row signature for this
window stage.
- for (String columnName : outputColumnNames) {
- int indexInRowSignature = rowSignature.indexOf(columnName);
- if (indexInRowSignature != -1 && bob.build().indexOf(columnName)
== -1) {
- ColumnType columnType =
rowSignature.getColumnType(indexInRowSignature).get();
- bob.add(columnName, columnType);
- log.info("Added column [%s] of type [%s] to row signature for
window stage.", columnName, columnType);
- } else {
- throw new ISE(
- "Found unexpected column [%s] already present in row
signature [%s].",
- columnName,
- rowSignature
- );
- }
+ /*
+ operatorList is a List<List<OperatorFactory>>, where each
List<OperatorFactory> corresponds to the operator factories
+ to be used for a different window stage.
+
+ We iterate over operatorList, and add the definition for a window stage
to QueryDefinitionBuilder.
+ */
+ for (int i = 0; i < operatorList.size(); i++) {
+ for (OperatorFactory operatorFactory : operatorList.get(i)) {
+ if (operatorFactory instanceof WindowOperatorFactory) {
+ List<String> outputColumnNames = ((WindowOperatorFactory)
operatorFactory).getProcessor().getOutputColumnNames();
+
+ // Need to add column names which are present in outputColumnNames
and rowSignature but not in bob,
+ // since they need to be present in the row signature for this
window stage.
+ for (String columnName : outputColumnNames) {
+ int indexInRowSignature = rowSignature.indexOf(columnName);
+ if (indexInRowSignature != -1 && bob.build().indexOf(columnName)
== -1) {
+ ColumnType columnType =
rowSignature.getColumnType(indexInRowSignature).get();
+ bob.add(columnName, columnType);
+ log.info("Added column [%s] of type [%s] to row signature for
window stage.", columnName, columnType);
+ } else {
+ throw new ISE(
+ "Found unexpected column [%s] already present in row
signature [%s].",
+ columnName,
+ rowSignature
+ );
}
}
}
+ }
- final RowSignature intermediateSignature = bob.build();
- final RowSignature stageRowSignature;
+ final RowSignature intermediateSignature = bob.build();
+ final RowSignature stageRowSignature;
- if (i + 1 == operatorList.size()) {
- stageRowSignature = finalWindowStageRowSignature;
- nextShuffleSpec = finalWindowStageShuffleSpec;
+ if (i + 1 == operatorList.size()) {
+ stageRowSignature = finalWindowStageRowSignature;
+ nextShuffleSpec = finalWindowStageShuffleSpec;
+ } else {
+ nextShuffleSpec = findShuffleSpecForNextWindow(operatorList.get(i +
1), maxWorkerCount);
Review Comment:
please open a separate PR and fix the stagebuilder rather than hacking it
backwards from here
--
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
To unsubscribe, e-mail: [email protected]
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
