[GitHub] [incubator-hudi] yihua commented on a change in pull request #1149: [WIP] [HUDI-472] Introduce configurations and new modes of sorting for bulk_insert
yihua commented on a change in pull request #1149: [WIP] [HUDI-472] Introduce
configurations and new modes of sorting for bulk_insert
URL: https://github.com/apache/incubator-hudi/pull/1149#discussion_r385909221
##
File path: hudi-client/src/main/java/org/apache/hudi/HoodieWriteClient.java
##
@@ -381,20 +384,30 @@ public static SparkConf registerClasses(SparkConf conf) {
}
}
+ private BulkInsertMapFunction getBulkInsertMapFunction(
Review comment:
For some subclasses in this PR, there are more complex logic in
`BulkInsertMapFunction`, e.g., `BulkInsertMapFunctionForNonSortedRecords`,
wondering if we should keep the API for now?
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.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-hudi] yihua commented on a change in pull request #1149: [WIP] [HUDI-472] Introduce configurations and new modes of sorting for bulk_insert
yihua commented on a change in pull request #1149: [WIP] [HUDI-472] Introduce
configurations and new modes of sorting for bulk_insert
URL: https://github.com/apache/incubator-hudi/pull/1149#discussion_r371038355
##
File path:
hudi-client/src/main/java/org/apache/hudi/func/bulkinsert/BulkInsertInternalPartitioner.java
##
@@ -0,0 +1,38 @@
+/*
+ * 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.hudi.func.bulkinsert;
+
+import org.apache.hudi.common.model.HoodieRecordPayload;
+import org.apache.hudi.table.UserDefinedBulkInsertPartitioner;
+
+public abstract class BulkInsertInternalPartitioner implements
+UserDefinedBulkInsertPartitioner {
Review comment:
@vinothchandar @n3nash @bvaradar Let's have a consensus on whether to change
the `UserDefinedBulkInsertPartitioner` interface. I'm fine with either way.
Right now I've also changed the interface so it will be a breaking change.
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.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-hudi] yihua commented on a change in pull request #1149: [WIP] [HUDI-472] Introduce configurations and new modes of sorting for bulk_insert
yihua commented on a change in pull request #1149: [WIP] [HUDI-472] Introduce
configurations and new modes of sorting for bulk_insert
URL: https://github.com/apache/incubator-hudi/pull/1149#discussion_r362302100
##
File path:
hudi-client/src/main/java/org/apache/hudi/table/UserDefinedBulkInsertPartitioner.java
##
@@ -31,4 +31,6 @@
public interface UserDefinedBulkInsertPartitioner {
JavaRDD> repartitionRecords(JavaRDD>
records, int outputSparkPartitions);
+
+ boolean arePartitionRecordsSorted();
Review comment:
Yeah, will do.
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.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-hudi] yihua commented on a change in pull request #1149: [WIP] [HUDI-472] Introduce configurations and new modes of sorting for bulk_insert
yihua commented on a change in pull request #1149: [WIP] [HUDI-472] Introduce configurations and new modes of sorting for bulk_insert URL: https://github.com/apache/incubator-hudi/pull/1149#discussion_r362302093 ## File path: hudi-client/src/main/java/org/apache/hudi/func/bulkinsert/NonSortPartitioner.java ## @@ -0,0 +1,38 @@ +/* + * 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.hudi.func.bulkinsert; + +import org.apache.hudi.common.model.HoodieRecord; +import org.apache.hudi.common.model.HoodieRecordPayload; +import org.apache.spark.api.java.JavaRDD; + +public class NonSortPartitioner Review comment: Yes, will rename. 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-hudi] yihua commented on a change in pull request #1149: [WIP] [HUDI-472] Introduce configurations and new modes of sorting for bulk_insert
yihua commented on a change in pull request #1149: [WIP] [HUDI-472] Introduce
configurations and new modes of sorting for bulk_insert
URL: https://github.com/apache/incubator-hudi/pull/1149#discussion_r362302090
##
File path:
hudi-client/src/main/java/org/apache/hudi/func/bulkinsert/BulkInsertInternalPartitioner.java
##
@@ -0,0 +1,38 @@
+/*
+ * 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.hudi.func.bulkinsert;
+
+import org.apache.hudi.common.model.HoodieRecordPayload;
+import org.apache.hudi.table.UserDefinedBulkInsertPartitioner;
+
+public abstract class BulkInsertInternalPartitioner implements
+UserDefinedBulkInsertPartitioner {
Review comment:
Yeah. I was also wondering about the custom implementation of
`UserDefinedBulkInsertPartitioner` in my previous comment..
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.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-hudi] yihua commented on a change in pull request #1149: [WIP] [HUDI-472] Introduce configurations and new modes of sorting for bulk_insert
yihua commented on a change in pull request #1149: [WIP] [HUDI-472] Introduce configurations and new modes of sorting for bulk_insert URL: https://github.com/apache/incubator-hudi/pull/1149#discussion_r362302051 ## File path: hudi-client/src/main/java/org/apache/hudi/func/CopyOnWriteInsertHandler.java ## @@ -0,0 +1,96 @@ +/* + * 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.hudi.func; + +import java.util.ArrayList; +import java.util.List; +import org.apache.hudi.WriteStatus; +import org.apache.hudi.common.model.HoodieRecord; +import org.apache.hudi.common.model.HoodieRecordPayload; +import org.apache.hudi.common.util.queue.BoundedInMemoryQueueConsumer; +import org.apache.hudi.config.HoodieWriteConfig; +import org.apache.hudi.func.CopyOnWriteLazyInsertIterable.HoodieInsertValueGenResult; +import org.apache.hudi.io.HoodieCreateHandle; +import org.apache.hudi.io.HoodieWriteHandle; +import org.apache.hudi.table.HoodieTable; + +/** + * Consumes stream of hoodie records from in-memory queue and writes to one or more create-handles. Review comment: Will do. 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-hudi] yihua commented on a change in pull request #1149: [WIP] [HUDI-472] Introduce configurations and new modes of sorting for bulk_insert
yihua commented on a change in pull request #1149: [WIP] [HUDI-472] Introduce
configurations and new modes of sorting for bulk_insert
URL: https://github.com/apache/incubator-hudi/pull/1149#discussion_r362302021
##
File path: hudi-client/src/main/java/org/apache/hudi/HoodieWriteClient.java
##
@@ -367,20 +370,30 @@ public static SparkConf registerClasses(SparkConf conf) {
}
}
+ private BulkInsertMapFunction getBulkInsertMapFunction(
+ boolean isSorted, String commitTime, HoodieWriteConfig config,
HoodieTable hoodieTable,
+ List fileIDPrefixes) {
+if (isSorted) {
+ return new BulkInsertMapFunctionForSortedRecords(
+ commitTime, config, hoodieTable, fileIDPrefixes);
+}
+return new BulkInsertMapFunctionForNonSortedRecords(
+commitTime, config, hoodieTable, fileIDPrefixes);
+ }
+
private JavaRDD bulkInsertInternal(JavaRDD>
dedupedRecords, String commitTime,
HoodieTable table, Option
bulkInsertPartitioner) {
Review comment:
I thought `UserDefinedBulkInsertPartitioner` is user-facing so I didn't
touch the interface. Looking at the source code again, I find that
`UserDefinedBulkInsertPartitioner` is only used here. Then we can just keep a
single one as `BulkInsertInternalPartitioner `.
BTW, is this interface (`UserDefinedBulkInsertPartitioner`) intended for
user to implement a custom partitioner? I don't seem to see a config to pass
in a custom implementation. wondering if we should provide such flexibility.
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.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
[GitHub] [incubator-hudi] yihua commented on a change in pull request #1149: [WIP] [HUDI-472] Introduce configurations and new modes of sorting for bulk_insert
yihua commented on a change in pull request #1149: [WIP] [HUDI-472] Introduce configurations and new modes of sorting for bulk_insert URL: https://github.com/apache/incubator-hudi/pull/1149#discussion_r362301880 ## File path: hudi-client/src/main/java/org/apache/hudi/config/HoodieWriteConfig.java ## @@ -77,6 +77,10 @@ private static final String DEFAULT_HOODIE_WRITE_STATUS_CLASS = WriteStatus.class.getName(); private static final String FINALIZE_WRITE_PARALLELISM = "hoodie.finalize.write.parallelism"; private static final String DEFAULT_FINALIZE_WRITE_PARALLELISM = DEFAULT_PARALLELISM; + private static final String BULKINSERT_SORT_ENABLED = "hoodie.bulkinsert.sort.enable"; Review comment: Yes, one config is better than two for the user to reason. I'll make the change accordingly. 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. For queries about this service, please contact Infrastructure at: us...@infra.apache.org With regards, Apache Git Services
[GitHub] [incubator-hudi] yihua commented on a change in pull request #1149: [WIP] [HUDI-472] Introduce configurations and new modes of sorting for bulk_insert
yihua commented on a change in pull request #1149: [WIP] [HUDI-472] Introduce
configurations and new modes of sorting for bulk_insert
URL: https://github.com/apache/incubator-hudi/pull/1149#discussion_r362275524
##
File path:
hudi-client/src/main/java/org/apache/hudi/func/bulkinsert/RDDPartitionLocalSortPartitioner.java
##
@@ -0,0 +1,44 @@
+/*
+ * 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.hudi.func.bulkinsert;
+
+import org.apache.hudi.common.model.HoodieRecord;
+import org.apache.hudi.common.model.HoodieRecordPayload;
+import org.apache.spark.HashPartitioner;
+import org.apache.spark.api.java.JavaRDD;
+import scala.Tuple2;
+
+public class RDDPartitionLocalSortPartitioner
+extends BulkInsertInternalPartitioner {
+
+ @Override
+ public JavaRDD> repartitionRecords(JavaRDD>
records,
+ int outputSparkPartitions) {
+return records.mapToPair(record ->
+new Tuple2<>(
+String.format("%s+%s", record.getPartitionPath(),
record.getRecordKey()), record))
+.repartitionAndSortWithinPartitions(new
HashPartitioner(outputSparkPartitions))
Review comment:
Yes, `RangePartitioner` is actually what I'm trying to look for.
For this specific Partitioner, it tries to avoid shuffling to speed up the
bulk insert but may introduce overlapping ranges. will check the side effects
of this.
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.
For queries about this service, please contact Infrastructure at:
us...@infra.apache.org
With regards,
Apache Git Services
