[HUDI-1656] Loading history data to new hudi table taking longer time

• Type: Improvement
• Status: Resolved
• Priority: Major
• Resolution: Fixed
• Fix Version/s: 0.7.0
• Labels: new-to-hudi sev:triage user-support-issues
• Assignee: Unassigned
• Reporter: Fredrick jose antony cruz
• Created: 04/Mar/21 05:13
• Updated: 16/Jan/22 06:29
• Resolved: 07/Oct/21 23:29

Description

spark-submit --jars /u/users/svcordrdats/order_hudi_poc/hudi-support-jars/org.apache.avro_avro-1.8.2.jar,/u/users/svcordrdats/order_hudi_poc/hudi-support-jars/spark-avro_2.11-2.4.4.jar,/u/users/svcordrdats/order_hudi_poc/hudi-support-jars/hudi-spark-bundle_2.11-0.7.0.jar --master yarn --deploy-mode cluster --num-executors 50 --executor-cores 4 --executor-memory 32g --driver-memory=24g --queue=default --conf spark.serializer=org.apache.spark.serializer.KryoSerializer --conf spark.driver.extraClassPath=org.apache.avro_avro-1.8.2.jar:spark-avro_2.11-2.4.4.jar --conf spark.executor.extraClassPath=org.apache.avro_avro-1.8.2.jar:spark-avro_2.11-2.4.4.jar:hudi-spark-bundle_2.11-0.7.0.jar --conf spark.memory.fraction=0.2 --driver-java-options "-XX:NewSize=1g -XX:SurvivorRatio=2 -XX:+UseCompressedOops -XX:+UseConcMarkSweepGC -XX:+UseParNewGC -XX:CMSInitiatingOccupancyFraction=70 -XX:+PrintGCDetails -XX:+PrintGCTimeStamps -XX:+PrintGCDateStamps -XX:+PrintGCApplicationStoppedTime -XX:+PrintGCApplicationConcurrentTime -XX:+PrintTenuringDistribution -XX:+HeapDumpOnOutOfMemoryError -XX:HeapDumpPath=/tmp/hoodie-heapdump.hprof" --files /usr/hdp/current/spark2-client/conf/hive-site.xml --class com.walmart.gis.order.workflows.WorkflowController lib/orders-poc-1.0.41-SNAPSHOT-shaded.jar workflow="stgStsWorkflow" runmode="global"

we are running on GCS cluster with 3 TB, 29 node cluster 870 v. core.

pom

<jdk-version>1.8</jdk-version>
<scala.version>2.11.12</scala.version>
<spark.version>2.3.0</spark.version>
<avro-version>1.8.2</avro-version>
<spark-avro.version>2.4.4</spark-avro.version>
<hudi-version>0.7.0</hudi-version>
<typesafe.config.version>1.4.0</typesafe.config.version>
<project.build.sourceEncoding>UTF-8</project.build.sourceEncoding>
<maven.compiler.target>1.8</maven.compiler.target>
<maven.compiler.source>1.8</maven.compiler.source>

stsDailyDf.write.format("org.apache.hudi")
.option("hoodie.cleaner.commits.retained", 2)
.option("hoodie.copyonwrite.record.size.estimate", 70)
.option("hoodie.parquet.small.file.limit", 100000000)
.option("hoodie.parquet.max.file.size", 128000000)
.option("hoodie.index.bloom.num_entries", 1800000)
.option("hoodie.bloom.index.filter.type", "DYNAMIC_V0")
.option("hoodie.bloom.index.filter.dynamic.max.entries", 2500000)
.option("hoodie.datasource.write.operation", "upsert")
.option("hoodie.datasource.write.storage.type", "COPY_ON_WRITE")
.option(DataSourceWriteOptions.RECORDKEY_FIELD_OPT_KEY, "sales_order_sts_line_key")
.option(DataSourceWriteOptions.PARTITIONPATH_FIELD_OPT_KEY, "REL_STS_DT")
.option(DataSourceWriteOptions.PRECOMBINE_FIELD_OPT_KEY, "src_upd_ts")
.option(HoodieWriteConfig.TABLE_NAME, tableName.toString)
.option("hoodie.bloom.index.bucketized.checking", "false")
.mode(SaveMode.Append)
.save(tablePath.toString)