#!/bin/sh # # # # # # # # # # # # # # # # # # # # # # # @author : jiangfan # @date : 2021-01-24 12:06:00 # # # # # # # # # # # # # # # # # # # # # # #!/usr/bin/env bash source ../../dmp_env.sh source ././../../ga_rawdata_analysis/common/tools.sh dt_today=$(date -d "$ScheduleTime 1 days ago" +"%Y%m%d") dt_two_days_ago=$(date -d "$ScheduleTime 2 days ago" +"%Y%m%d") seven_days_ago=$(date -d "$ScheduleTime 7 days ago" +"%Y%m%d") fifteen_days_ago=$(date -d "$ScheduleTime 15 days ago" +"%Y%m%d") dt_dash_tow_days=$(date -d "$ScheduleTime 2 days ago" +"%Y-%m-%d") dt_dash_one_days=$(date -d "$ScheduleTime 1 days ago" +"%Y-%m-%d") dt_slash_today=$(date -d "$ScheduleTime 1 days ago" +"%Y/%m/%d") year=${dt_today:0:4} month=${dt_today:4:2} day=${dt_today:6:2} hh="23" check_await "${UPARPU_PLUGIN_QCC_PACKAGE}/yyyy=${year}/mm=${month}/dd=${day}" #检查表分区是否已经存在,不存在等待五个小时退出脚本 partition_info="dt='${dt_two_days_ago}',business='btop'" check_await_hive_partition "dwh.dm_install_list_v2 " $partition_info 18000 partition_info="yyyy='${year}',mm='${month}',dd='${day}',hh='${hh}'" check_await_hive_partition "uparpu_main.uparpu_strategy_app_v2" $partition_info 18000 sleep 50 OUTPUT_PATH="${BTOP_DAILY_PATH}/${dt_slash_today}" spark-submit --class mobvista.dmp.datasource.btop.BtopDaily \ --conf spark.network.timeout=720s \ --conf spark.driver.maxResultSize=4g \ --conf spark.default.parallelism=3000 \ --conf spark.sql.shuffle.partitions=3000 \ --conf spark.sql.broadcastTimeout=1200 \ --conf spark.sql.autoBroadcastJoinThreshold=31457280 \ --deploy-mode cluster --executor-memory 6g --driver-memory 4g --executor-cores 2 --num-executors 60 \ ../../${JAR} \ -output ${OUTPUT_PATH} -coalesce 200 -dt_today ${dt_today} -dt_dash_tow_days ${dt_dash_tow_days} -seven_days_ago ${seven_days_ago} -fifteen_days_ago ${fifteen_days_ago} \ -dt_two_days_ago ${dt_two_days_ago} -dt_dash_one_days ${dt_dash_one_days} if [[ $? -ne 0 ]];then exit 255 fi mount_partition "etl_btop_daily" "dt='${dt_today}'" "$OUTPUT_PATH" hadoop fs -touchz ${OUTPUT_PATH}/_SUCCESS