#!/bin/bash # # # # # # # # # # # # # # # # # # # # # # # @author :fengliang # @revision:2017-09-01 # # # # # # # # # # # # # # # # # # # # # # source ../../dmp_env.sh LOG_TIME=$(date +%Y%m%d -d "-1 day $ScheduleTime") date_path=$(date +%Y/%m/%d -d "-1 day $ScheduleTime") INPUT_PATH="${DM_INSTALL_LIST}/$date_path/adn_install" OUTPUT_PATH="${DM_INSTALL_LIST}_v2/$date_path/adn_install" check_await ${INPUT_PATH}/_SUCCESS hadoop fs -rm -r ${OUTPUT_PATH} REDUCE_NUM=$(calculate_reduce_num ${INPUT_PATH}) hadoop jar ../../${JAR} mobvista.dmp.main.ParseInstallRCFile \ -Dmapreduce.fileoutputcommitter.algorithm.version=2 \ $INPUT_PATH $OUTPUT_PATH ${REDUCE_NUM} if [ $? -ne 0 ];then exit 255 fi : ' business="adn_install" spark-submit --class mobvista.dmp.datasource.dm.DmInstallListOrc \ --name "mobvista.dmp.datasource.dm.DmInstallListOrc.${business}" \ --conf spark.yarn.executor.memoryOverhead=2048 \ --conf spark.default.parallelism=20 \ --conf spark.speculation=true \ --conf spark.speculation.quantile=0.8 \ --conf spark.speculation.multiplier=1.2 \ --files ${HIVE_SITE_PATH} \ --jars /data/hadoop-alternative/hive/auxlib/Common-SerDe-1.0-SNAPSHOT.jar \ --master yarn --deploy-mode cluster --executor-memory 6g --driver-memory 4g --executor-cores 2 --num-executors 10 \ ../../${JAR} \ -output ${OUTPUT_PATH} -input ${INPUT_PATH} -business ${business} ' if [ $? -ne 0 ];then exit 255 fi mount_partition "dm_install_list_v2" "dt='$LOG_TIME', business='adn_install'" "$OUTPUT_PATH" if [ $? -ne 0 ];then exit 255 fi