#!/usr/bin/env bash source ../dmp_env.sh today=$(date -d "$ScheduleTime 1 days ago" +"%Y/%m/%d") yesterday=$(date -d "$ScheduleTime 2 days ago" +"%Y/%m/%d") dt_today=$(date -d "$ScheduleTime 1 days ago" +"%Y%m%d") dt_yesterday=$(date -d "$ScheduleTime 2 days ago" +"%Y%m%d") echo ${today} echo ${yesterday} : ' #由日期获得上周日日期 week=`date -d "$today" +%w` echo "week=$week" if [ "$week" -eq "0" ];then week=7 #若为周日,则表示为7 fi if [ "$week" -eq "1" ];then week=8 #若为周一,则表示为8,取上上周日数据 fi last_sunday=$(date +%Y%m%d -d "-$week day $today") echo "last_sunday=$last_sunday" ' OUTPUT_PATH="${ETL_3S_POSTBACK_DAILY_PATH}/${today}" POSTBACK_INSTALL_3S="${POSTBACK_INSTALL_3S_PATH}/${today}/virginia/23/_SUCCESS" POSTBACK_EVENT_3S="${POSTBACK_EVENT_3S_PATH}/${today}/virginia/23/_SUCCESS" check_await "${POSTBACK_INSTALL_3S}" check_await "${POSTBACK_EVENT_3S}" check_await "${DEVICE_ID_MD5_MATCH_PATH}/${yesterday}/_SUCCESS" hadoop fs -rm -r "$OUTPUT_PATH/" spark-submit --class mobvista.dmp.datasource.postback_3s.PostBackDaily \ --conf spark.network.timeout=720s \ --conf spark.default.parallelism=2000 \ --conf spark.sql.shuffle.partitions=1000 \ --conf spark.sql.broadcastTimeout=1200 \ --conf spark.sql.autoBroadcastJoinThreshold=31457280 \ --jars s3://mob-emr-test/dataplatform/DataWareHouse/offline/myjar/hive-hcatalog-core-2.3.3.jar \ --master yarn --deploy-mode cluster --executor-memory 6g --driver-memory 4g --executor-cores 3 --num-executors 100 \ ../${JAR} -output ${OUTPUT_PATH} -coalesce 100 \ -today ${dt_today} -last_sunday ${dt_yesterday} if [ $? -ne 0 ];then exit 255 fi mount_partition "etl_3s_postback_daily" "dt='${dt_today}'" "$OUTPUT_PATH" hadoop fs -touchz ${OUTPUT_PATH}/_SUCCESS