#!/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}/$date_path/clever" OUTPUT_PATH="${DM_INSTALL_LIST}_v2/$date_path/clever" check_await ${INPUT_PATH}/_SUCCESS hadoop fs -rm -r $OUTPUT_PATH hadoop jar ../${JAR} mobvista.dmp.main.ParseInstallRCFile \ -Dmapreduce.fileoutputcommitter.algorithm.version=2 \ $INPUT_PATH $OUTPUT_PATH 20 if [ $? -ne 0 ];then exit 255 fi : ' business="clever" spark-submit --class mobvista.dmp.datasource.dm.DmInstallListOrc \ --name "mobvista.dmp.datasource.dm.DmInstallListOrc.${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 \ --deploy-mode cluster --executor-memory 10g --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='clever'" "$OUTPUT_PATH" if [ $? -ne 0 ];then exit 255 fi