You are viewing a plain text version of this content. The canonical link for it is here.
Posted to commits@druid.apache.org by GitBox <gi...@apache.org> on 2021/11/03 09:09:45 UTC
[GitHub] [druid] abhishekagarwal87 commented on a change in pull request #11848: Add support for multi dimension range partitioning
abhishekagarwal87 commented on a change in pull request #11848:
URL: https://github.com/apache/druid/pull/11848#discussion_r741708297
##########
File path: indexing-service/src/main/java/org/apache/druid/indexing/common/task/batch/parallel/PartialRangeSegmentGenerateTask.java
##########
@@ -92,20 +92,20 @@ public PartialRangeSegmentGenerateTask(
this.intervalToPartitions = intervalToPartitions;
}
- private static String getPartitionDimension(ParallelIndexIngestionSpec ingestionSpec)
+ private static List<String> getPartitionDimensions(ParallelIndexIngestionSpec ingestionSpec)
{
PartitionsSpec partitionsSpec = ingestionSpec.getTuningConfig().getPartitionsSpec();
Preconditions.checkArgument(
- partitionsSpec instanceof SingleDimensionPartitionsSpec,
+ partitionsSpec instanceof MultiDimensionPartitionsSpec,
"%s partitionsSpec required",
Review comment:
it should rather say that either single_dim or multi_dim partition spec is required
##########
File path: indexing-service/src/main/java/org/apache/druid/indexing/common/task/batch/parallel/distribution/ArrayOfStringTuplesSerDe.java
##########
@@ -0,0 +1,82 @@
+/*
+ * 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.indexing.common.task.batch.parallel.distribution;
+
+import org.apache.datasketches.ArrayOfItemsSerDe;
+import org.apache.datasketches.ArrayOfStringsSerDe;
+import org.apache.datasketches.memory.Memory;
+import org.apache.datasketches.memory.WritableMemory;
+import org.apache.datasketches.memory.internal.UnsafeUtil;
+import org.apache.druid.data.input.StringTuple;
+
+/**
+ * Serde for {@link StringTuple}.
+ *
+ * Implementation similar to {@link ArrayOfStringsSerDe}.
+ */
+public class ArrayOfStringTuplesSerDe extends ArrayOfItemsSerDe<StringTuple>
+{
+ private static final ArrayOfStringsSerDe STRINGS_SERDE = new ArrayOfStringsSerDe();
+
+ @Override
+ public byte[] serializeToByteArray(StringTuple[] items)
+ {
+ int length = 0;
+ final byte[][] itemsBytes = new byte[items.length][];
+ for (int i = 0; i < items.length; i++) {
+ itemsBytes[i] = STRINGS_SERDE.serializeToByteArray(items[i].toArray());
+ length += itemsBytes[i].length + Integer.BYTES;
+ length += items[i].size() + Integer.BYTES;
Review comment:
I think it should be `length += Integer.BYTES`;
##########
File path: indexing-service/src/main/java/org/apache/druid/indexing/common/task/batch/parallel/iterator/RangePartitionIndexTaskInputRowIteratorBuilder.java
##########
@@ -79,36 +80,46 @@ public HandlingInputRowIterator build()
}
private static HandlingInputRowIterator.InputRowHandler createOnlySingleDimensionValueRowsHandler(
- String partitionDimension
+ List<String> partitionDimensions
)
{
- return inputRow -> {
- int dimensionValueCount = getSingleOrNullDimensionValueCount(inputRow, partitionDimension);
- return dimensionValueCount != 1;
- };
+ return inputRow -> isRowHandled(inputRow, partitionDimensions, dimValueCount -> dimValueCount != 1);
}
private static HandlingInputRowIterator.InputRowHandler createOnlySingleOrNullDimensionValueRowsHandler(
- String partitionDimension
+ List<String> partitionDimensions
)
{
- return inputRow -> {
- int dimensionValueCount = getSingleOrNullDimensionValueCount(inputRow, partitionDimension);
- return dimensionValueCount > 1; // Rows.objectToStrings() returns an empty list for a single null value
- };
+ // Rows.objectToStrings() returns an empty list for a single null value
+ return inputRow -> isRowHandled(inputRow, partitionDimensions, dimValueCount -> dimValueCount > 1);
}
- private static int getSingleOrNullDimensionValueCount(InputRow inputRow, String partitionDimension)
+ /**
+ * @param valueCountPredicate Predicate that must be satisfied
+ * for atleast one of the partitionDimensions for the row to be marked as handled.
+ * @return true when the given InputRow should be marked handled
+ * and need not be processed further.
+ */
+ private static boolean isRowHandled(
+ InputRow inputRow,
+ List<String> partitionDimensions,
+ Predicate<Integer> valueCountPredicate
+ )
{
- List<String> dimensionValues = inputRow.getDimension(partitionDimension);
- int dimensionValueCount = dimensionValues.size();
- if (dimensionValueCount > 1) {
- throw new IAE(
- "Cannot partition on multi-value dimension [%s] for input row [%s]",
- partitionDimension,
- inputRow
- );
+ for (String dimension : partitionDimensions) {
+ int dimensionValueCount = inputRow.getDimension(dimension).size();
+ if (dimensionValueCount > 1) {
+ throw new IAE(
+ "Cannot partition on multi-value dimension [%s] for input row [%s]",
+ dimension,
+ inputRow
+ );
+ } else if (valueCountPredicate.test(dimensionValueCount)) {
Review comment:
ok. there is one more value predicate. thinking more on it, `inputRow -> isRowHandled(inputRow, partitionDimensions, dimValueCount -> dimValueCount > 1)` can very well be inputRow -> isRowHandled(inputRow, partitionDimensions, dimValueCount -> false)`. Just trying to understand this logic better.
##########
File path: core/src/main/java/org/apache/druid/indexer/partitions/MultiDimensionPartitionsSpec.java
##########
@@ -0,0 +1,199 @@
+/*
+ * 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.indexer.partitions;
+
+
+import com.fasterxml.jackson.annotation.JsonCreator;
+import com.fasterxml.jackson.annotation.JsonIgnore;
+import com.fasterxml.jackson.annotation.JsonProperty;
+import com.google.common.base.Preconditions;
+import org.apache.druid.indexer.Property;
+
+import javax.annotation.Nullable;
+import javax.validation.constraints.NotNull;
+import java.util.List;
+import java.util.Objects;
+
+/**
+ * Range partition a segment by multiple dimensions.
+ */
+public class MultiDimensionPartitionsSpec implements DimensionBasedPartitionsSpec
+{
+ public static final String NAME = "multi_dim";
+
+ private final Integer targetRowsPerSegment;
+ private final Integer maxRowsPerSegment;
+ private final List<String> partitionDimensions;
+ private final boolean assumeGrouped;
Review comment:
what does assumeGrouped mean?
##########
File path: indexing-service/src/main/java/org/apache/druid/indexing/common/task/batch/parallel/iterator/RangePartitionIndexTaskInputRowIteratorBuilder.java
##########
@@ -79,36 +80,46 @@ public HandlingInputRowIterator build()
}
private static HandlingInputRowIterator.InputRowHandler createOnlySingleDimensionValueRowsHandler(
- String partitionDimension
+ List<String> partitionDimensions
)
{
- return inputRow -> {
- int dimensionValueCount = getSingleOrNullDimensionValueCount(inputRow, partitionDimension);
- return dimensionValueCount != 1;
- };
+ return inputRow -> isRowHandled(inputRow, partitionDimensions, dimValueCount -> dimValueCount != 1);
}
private static HandlingInputRowIterator.InputRowHandler createOnlySingleOrNullDimensionValueRowsHandler(
- String partitionDimension
+ List<String> partitionDimensions
)
{
- return inputRow -> {
- int dimensionValueCount = getSingleOrNullDimensionValueCount(inputRow, partitionDimension);
- return dimensionValueCount > 1; // Rows.objectToStrings() returns an empty list for a single null value
- };
+ // Rows.objectToStrings() returns an empty list for a single null value
+ return inputRow -> isRowHandled(inputRow, partitionDimensions, dimValueCount -> dimValueCount > 1);
}
- private static int getSingleOrNullDimensionValueCount(InputRow inputRow, String partitionDimension)
+ /**
+ * @param valueCountPredicate Predicate that must be satisfied
+ * for atleast one of the partitionDimensions for the row to be marked as handled.
+ * @return true when the given InputRow should be marked handled
+ * and need not be processed further.
+ */
+ private static boolean isRowHandled(
+ InputRow inputRow,
+ List<String> partitionDimensions,
+ Predicate<Integer> valueCountPredicate
+ )
{
- List<String> dimensionValues = inputRow.getDimension(partitionDimension);
- int dimensionValueCount = dimensionValues.size();
- if (dimensionValueCount > 1) {
- throw new IAE(
- "Cannot partition on multi-value dimension [%s] for input row [%s]",
- partitionDimension,
- inputRow
- );
+ for (String dimension : partitionDimensions) {
+ int dimensionValueCount = inputRow.getDimension(dimension).size();
+ if (dimensionValueCount > 1) {
+ throw new IAE(
+ "Cannot partition on multi-value dimension [%s] for input row [%s]",
+ dimension,
+ inputRow
+ );
+ } else if (valueCountPredicate.test(dimensionValueCount)) {
Review comment:
so the `valueCountPredicate` is `dimValueCount -> dimValueCount > 1` but if dimensionValueCount is greater than 1, we will throw an exception. so this should always return false. isn't it?
--
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: commits-unsubscribe@druid.apache.org
For queries about this service, please contact Infrastructure at:
users@infra.apache.org
---------------------------------------------------------------------
To unsubscribe, e-mail: commits-unsubscribe@druid.apache.org
For additional commands, e-mail: commits-help@druid.apache.org