uc_other_data_to_dmp_v2.sh 2.28 KB
#!/usr/bin/env bash

source ../../dmp_env.sh
source ././../../ga_rawdata_analysis/common/tools.sh

dt_today=$(date -d "$ScheduleTime" +"%Y%m%d")
dt_oneday_ago=$(date -d "$ScheduleTime 1 days ago" +"%Y%m%d")
dt_slash_today=$(date -d "$ScheduleTime" +"%Y/%m/%d")
update=$(date -d "$ScheduleTime" +"%Y-%m-%d")

dt_slash_oneday=$(date -d "$ScheduleTime 1 days ago" +"%Y/%m/%d")
expire_date=$(date -d "$ScheduleTime 61 days ago" +"%Y%m%d")
expire_date_path=$(date -d "$ScheduleTime 61 days ago" +"%Y/%m/%d")
check_await "${DM_INSTALL_LIST}_v2/${dt_slash_oneday}/dsp_req/_SUCCESS"

DMP_INSTALL_LIST_OUTPUT_PATH="${DM_INSTALL_LIST}_v2/${dt_slash_today}/uc_activation"

EXPIRE_OUTPUT_PATH="${UC_LAHUO_TMP_DAILY_TO_S3}/${expire_date_path}/uc_activation_other_data"

# OUTPUT_PATH01="${UC_LAHUO_TMP_DAILY_TO_S3}/${dt_slash_today}/uc_activation_other_data/4b5a58_ucbes"
# OUTPUT_PATH02="${UC_LAHUO_TMP_DAILY_TO_S3}/${dt_slash_today}/uc_activation_other_data/d3f521_ucbes"
# OUTPUT_PATH03="${UC_LAHUO_TMP_DAILY_TO_S3}/${dt_slash_today}/uc_activation_other_data/4b5a58_ucoppo"
# OUTPUT_PATH04="${UC_LAHUO_TMP_DAILY_TO_S3}/${dt_slash_today}/uc_activation_other_data/d3f521_ucoppo"

OUTPUT_PATH="${UC_LAHUO_TMP_DAILY_TO_S3}/${dt_slash_today}/uc_activation_other_data"

hadoop fs -rm -r "${OUTPUT_PATH}"

spark-submit --class mobvista.dmp.datasource.taobao.UCOtherDataToDmpV2 \
  --conf spark.network.timeout=720s \
  --conf spark.default.parallelism=3000 \
  --conf spark.sql.shuffle.partitions=3000 \
  --conf spark.sql.broadcastTimeout=1200 \
  --conf spark.sql.autoBroadcastJoinThreshold=31457280 \
  --files ${HIVE_SITE_PATH} \
  --master yarn --deploy-mode cluster --executor-memory 10g --driver-memory 6g --executor-cores 5 --num-executors 120 \
  ../../${JAR} \
  -output ${OUTPUT_PATH} \
  -dt_today ${dt_today} -dt_oneday_ago ${dt_oneday_ago} -update ${update}

if [ $? -ne 0 ]; then
  exit 255
fi

mount_partition "uc_lahuo_tmp_daily_to_s3" "dt='${dt_today}', business='uc_activation_other_data'" "${OUTPUT_PATH}"

unmount_partition "uc_lahuo_tmp_daily_to_s3" "dt='${expire_date}', business='uc_activation_other_data'" "${EXPIRE_OUTPUT_PATH}"

HIVE_CMD=$(hive_func)

$HIVE_CMD -v -hivevar dt_today ${dt_today} -f uc_other_data_to_dmp.sql

if [ $? -ne 0 ]; then
  exit 255
fi

hadoop fs -touchz ${DMP_INSTALL_LIST_OUTPUT_PATH}/_SUCCESS