#!/bin/bash # # # # # # # # # # # # # # # # # # # # # # # @author :wangjf # @revision:2019-08-28 17:36:39 # # # # # # # # # # # # # # # # # # # # # # source ../dmp_env.sh LOG_TIME=$(date +%Y%m%d -d "$ScheduleTime") date_path=$(date +%Y/%m/%d -d "$ScheduleTime") business="ali" INPUT_PATH="${DM_INSTALL_LIST}/$date_path/${business}" OUTPUT_PATH="${DM_INSTALL_LIST}_v2/$date_path/${business}" check_await ${INPUT_PATH}/_SUCCESS hadoop fs -rm -r ${OUTPUT_PATH} REDUCE_NUM=$(calculate_reduce_num ${INPUT_PATH}) export HADOOP_USER_CLASSPATH_FIRST=yes export HADOOP_CLASSPATH="${HADOOP_CLASSPATH}:/data/hadoop-alternative/hive/lib/*" hadoop jar ../${JAR} mobvista.dmp.main.ParseInstallRCFile \ -Dmapreduce.fileoutputcommitter.algorithm.version=2 \ ${INPUT_PATH} ${OUTPUT_PATH} 100 : ' spark-submit --class mobvista.dmp.datasource.dm.DmInstallListOrc \ --name "mobvista.dmp.datasource.dm.DmInstallListOrc.${business}" \ --conf spark.yarn.executor.memoryOverhead=2048 \ --conf spark.default.parallelism=500 \ --conf spark.speculation=true \ --conf spark.speculation.quantile=0.8 \ --conf spark.speculation.multiplier=1.2 \ --files ${HIVE_SITE_PATH} \ --jars /data/hadoop-alternative/hive/auxlib/Common-SerDe-1.0-SNAPSHOT.jar \ --master yarn --deploy-mode cluster --executor-memory 10g --driver-memory 4g --executor-cores 3 --num-executors 80 \ ../${JAR} \ -output ${OUTPUT_PATH} -input ${INPUT_PATH} -business ${business} if [ $? -ne 0 ];then exit 255 fi ' unset HADOOP_CLASSPATH mount_partition "dm_install_list_v2" "dt='$LOG_TIME', business='${business}'" "$OUTPUT_PATH" if [[ $? -ne 0 ]];then exit 255 fi