#!/usr/bin/env bash source ../../dmp_env.sh source ././../../ga_rawdata_analysis/common/tools.sh echo "job begin!!!" dt_today=$(date -d "$ScheduleTime" +"%Y%m%d") dt_slash_today=$(date -d "$ScheduleTime" +"%Y/%m/%d") update=$(date -d "$ScheduleTime" +"%Y-%m-%d") dt_oneday_ago=$(date -d "$ScheduleTime 1 days ago" +"%Y/%m/%d") UC_IMEIMD5_REQUEST_INPUT_PATH="${UC_LAHUO_DAILY_PATH}/${dt_oneday_ago}/imeimd5_request_data" UC_OAIDMD5_REQUEST_INPUT_PATH="${UC_LAHUO_DAILY_PATH}/${dt_oneday_ago}/oaidmd5_request_data" IMEIMD5_RESPONSE_INPUT_PATH="${UC_LAHUO_DAILY_TMP_PATH}/${dt_slash_today}/imeimd5" OAIDMD5_RESPONSE_INPUT_PATH="${UC_LAHUO_DAILY_TMP_PATH}/${dt_slash_today}/oaidmd5" IMEIMD5_OUTPUT="${UC_LAHUO_TMP_DAILY_TO_S3}/${dt_slash_today}/imeimd5/foractivation" OAIDMD5_OUTPUT="${UC_LAHUO_TMP_DAILY_TO_S3}/${dt_slash_today}/oaidmd5/foractivation" IMEIMD5_NOT_ACTIVATION_OUTPUT="${UC_LAHUO_TMP_DAILY_TO_S3}/${dt_slash_today}/imeimd5/notforactivation" OAIDMD5_NOT_ACTIVATION_OUTPUT="${UC_LAHUO_TMP_DAILY_TO_S3}/${dt_slash_today}/oaidmd5/notforactivation" check_await "${IMEIMD5_RESPONSE_INPUT_PATH}/_SUCCESS" check_await "${OAIDMD5_RESPONSE_INPUT_PATH}/_SUCCESS" hadoop fs -rm -r "${IMEIMD5_OUTPUT}" hadoop fs -rm -r "${OAIDMD5_OUTPUT}" hadoop fs -rm -r "${IMEIMD5_NOT_ACTIVATION_OUTPUT}" hadoop fs -rm -r "${OAIDMD5_NOT_ACTIVATION_OUTPUT}" spark-submit --class mobvista.dmp.datasource.taobao.UCTmpDataToDMP \ --conf spark.network.timeout=720s \ --conf spark.default.parallelism=2000 \ --conf spark.sql.shuffle.partitions=2000 \ --conf spark.sql.broadcastTimeout=1200 \ --conf spark.yarn.executor.memoryOverhead=4096 \ --conf spark.sql.autoBroadcastJoinThreshold=31457280 \ --files ${HIVE_SITE_PATH} \ --master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 4 --num-executors 40 \ ../../${JAR} -imeiRequestInput ${UC_IMEIMD5_REQUEST_INPUT_PATH} -oaidRequestInput ${UC_OAIDMD5_REQUEST_INPUT_PATH} \ -imeiResponseInput ${IMEIMD5_RESPONSE_INPUT_PATH} -oaidResponseInput ${OAIDMD5_RESPONSE_INPUT_PATH} \ -imeiOutput ${IMEIMD5_OUTPUT} -oaidOutput ${OAIDMD5_OUTPUT} \ -imeiNotActivationOutput ${IMEIMD5_NOT_ACTIVATION_OUTPUT} -oaidNotActivationOutput ${OAIDMD5_NOT_ACTIVATION_OUTPUT} \ -update ${update} if [ $? -ne 0 ];then exit 255 fi DMP_INSTALL_LIST_OUTPUT_PATH="${DM_INSTALL_LIST}_v2/${dt_slash_today}/uc_activation" mount_partition "dm_install_list_v2" "dt='${dt_today}', business='uc_activation'" "${DMP_INSTALL_LIST_OUTPUT_PATH}" mount_partition "uc_lahuo_tmp_daily_to_s3" "dt='${dt_today}', business='uc_activation'" "${UC_LAHUO_TMP_DAILY_TO_S3}/${dt_slash_today}" HIVE_CMD=$(hive_func) $HIVE_CMD -v -hivevar dt_today ${dt_today} -f uc_lahuo_data_to_dmp.sql if [ $? -ne 0 ];then exit 255 fi #hadoop fs -touchz ${DMP_INSTALL_LIST_OUTPUT_PATH}/_SUCCESS echo "good job"