Commit 381415e2 by wang-jinfeng

增大 driver-memory,避免出现 driver pod OOMKilled

parent 5c0515cb
......@@ -35,7 +35,7 @@ spark-submit --class mobvista.dmp.datasource.newtag.MatchInterestTag \
--conf spark.yarn.executor.memoryOverhead=3072 \
--files ${HIVE_SITE_PATH} \
--jars /data/hadoop-alternative/hive/auxlib/Common-SerDe-1.0-SNAPSHOT.jar \
--master yarn --deploy-mode cluster --executor-memory 6g --driver-memory 5g --executor-cores 2 --num-executors 80 \
--master yarn --deploy-mode cluster --executor-memory 6g --driver-memory 4g --executor-cores 2 --num-executors 80 \
../../${JAR} \
-date $date -manualOutput ${output_path} -business ${business} -storeOutput ${store_output_path} -coalesce 2000
......
......@@ -28,7 +28,7 @@ spark-submit --class mobvista.dmp.datasource.adn_request_other.AdnRequestOtherIn
--conf spark.yarn.executor.memoryOverhead=2048 \
--conf spark.sql.shuffle.partitions=2000 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 6g --driver-memory 3g --executor-cores 2 --num-executors 200 \
--master yarn --deploy-mode cluster --executor-memory 6g --driver-memory 4g --executor-cores 2 --num-executors 200 \
../../${JAR} \
-input "${INPUT_PATH}" -output $OUTPUT_PATH -date $date -oldInput $OLD_INPUT_PATH -parallelism 2000 -coalesce 2000
if [ $? -ne 0 ]; then
......
......@@ -29,7 +29,7 @@ spark-submit --class mobvista.dmp.datasource.adn_adx.AdnAdxDeviceTag \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--jars s3://mob-emr-test/dataplatform/DataWareHouse/offline/myjar/hive-hcatalog-core-2.3.3.jar \
--master yarn --deploy-mode cluster --name AdnAdxDeviceTag --executor-memory 4g --driver-memory 2g --executor-cores 2 --num-executors 32 \
--master yarn --deploy-mode cluster --name AdnAdxDeviceTag --executor-memory 4g --driver-memory 4g --executor-cores 2 --num-executors 32 \
../${JAR} -outputadxdevtag ${OUTPUT_ADN_ADX_DEVICE_TAG_PATH} \
-coalesce 80 \
-today ${dt_today} -yesterday ${dt_yesterday}
......
......@@ -60,7 +60,7 @@ spark-submit --class mobvista.dmp.datasource.age_gender.GetAgeGender \
--conf spark.executor.extraJavaOptions="-XX:+UseG1GC" \
--files ${HIVE_SITE_PATH} \
--jars ${JARS} \
--master yarn --deploy-mode cluster --executor-memory 6g --driver-memory 2g --executor-cores 2 --num-executors 50 \
--master yarn --deploy-mode cluster --executor-memory 6g --driver-memory 4g --executor-cores 2 --num-executors 50 \
../${JAR} -ageOutput ${AGE_OUTPUT_PATH} -genderOutput ${GENDER_OUTPUT_PATH} -date ${GA_TOTAL_DATE} -business ${business}
if [[ $? -ne 0 ]];then
......
......@@ -28,7 +28,7 @@ spark-submit --class mobvista.dmp.datasource.TO.TODaily \
--conf spark.sql.broadcastTimeout=1200 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 2g --executor-cores 6 --num-executors 20 \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 6 --num-executors 20 \
../../${JAR} \
-output ${OUTPUT_PATH} -coalesce 200 -dt_dash_today ${dt_dash_today}
......
......@@ -35,7 +35,7 @@ spark-submit --class mobvista.dmp.datasource.taobao.EtlAliIosActivitionDaily \
--conf spark.yarn.executor.memoryOverhead=4096 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 9g --driver-memory 3g --executor-cores 4 --num-executors 60 \
--master yarn --deploy-mode cluster --executor-memory 10g --driver-memory 4g --executor-cores 4 --num-executors 60 \
../../${JAR} -output ${OUTPUT_PATH} -outputdaily ${ALI_OUTPUT_DAILY_PATH} -coalesce 500 \
-yesterday ${yesterday} -today ${dt_today} -dt_dash_today ${dt_dash_today} -dt_dash_rec14day ${dt_dash_rec14day} \
-request_count_result "${IOS_REQUEST_COUNT_RESULT}/${dt_today}" -last_req_day ${last_req_day}
......
......@@ -30,7 +30,7 @@ spark-submit --class mobvista.dmp.datasource.taobao.EtlAliOaidActivitionDaily \
--conf spark.yarn.executor.memoryOverhead=4096 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 9g --driver-memory 3g --executor-cores 4 --num-executors 60 \
--master yarn --deploy-mode cluster --executor-memory 10g --driver-memory 4g --executor-cores 4 --num-executors 60 \
../../${JAR} -output ${OUTPUT_PATH} -outputdaily ${ALI_OAID_OUTPUT_DAILY_PATH} -coalesce 500 \
-yesterday ${yesterday} -today ${dt_today} -dt_dash_today ${dt_dash_today} -dt_dash_rec14day ${dt_dash_rec14day} \
-request_count_result "${OAID_REQUEST_COUNT_RESULT}/${dt_today}" -last_req_day ${last_req_day}
......
......@@ -34,7 +34,7 @@ spark-submit --class mobvista.dmp.datasource.taobao.EtlAliActivitionDaily \
--conf spark.yarn.executor.memoryOverhead=4096 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 9g --driver-memory 3g --executor-cores 4 --num-executors 60 \
--master yarn --deploy-mode cluster --executor-memory 10g --driver-memory 4g --executor-cores 4 --num-executors 60 \
../../${JAR} -output ${OUTPUT_PATH} -outputdaily ${ALI_OUTPUT_DAILY_PATH} -coalesce 500 \
-yesterday ${yesterday} -today ${dt_today} -dt_dash_today ${dt_dash_today} -dt_dash_rec14day ${dt_dash_rec14day} \
-request_count_result "${REQUEST_COUNT_RESULT}/${dt_today}" -last_req_day ${last_req_day}
......
......@@ -85,7 +85,7 @@ spark-submit --class mobvista.dmp.datasource.taobao.EtlAliActivitionPostBackDail
--conf spark.sql.broadcastTimeout=1200 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 2g --executor-cores 3 --num-executors 60 \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 3 --num-executors 60 \
../${JAR} -output ${OUTPUT_PATH} -iosoutput ${ALI_IOS_OUTPUT} -oaidoutput ${ALI_OAID_OUTPUT} -coalesce 50 \
-today ${dt_today} -update_date ${dt_dash_today} \
-dt_dash_rec15day ${dt_dash_rec15day} -syn_to_3s ${ALI_USER_ACTIVATION_SYS_TO3S_PATH} -syn_3s_day ${syn_3s_day}
......
......@@ -62,7 +62,7 @@ spark-submit --class mobvista.dmp.datasource.taobao.EtlAliActivitionPostBackDail
--conf spark.sql.broadcastTimeout=1200 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 3g --executor-cores 6 --num-executors 70 \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 6 --num-executors 70 \
../../${JAR} -output ${OUTPUT_PATH} -iosoutput ${ALI_IOS_OUTPUT} -oaidoutput ${ALI_OAID_OUTPUT} -coalesce 300 \
-today ${dt_today} -update_date ${dt_dash_today} -dt_taobao_postback_day ${dt_taobao_postback_day}\
-dt_dash_rec15day ${dt_dash_rec15day} -syn_to_3s ${ALI_USER_ACTIVATION_SYS_TO3S_PATH} -syn_3s_day ${syn_3s_day}
......
......@@ -30,7 +30,7 @@ spark-submit --class mobvista.dmp.datasource.taobao.EtlH18FromDmInstallListV2 \
--conf spark.sql.broadcastTimeout=1200 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 2g --executor-cores 6 --num-executors 60 \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 6 --num-executors 60 \
../../${JAR} \
-h18_imei ${IMEI_H_18_GUANGDIANTONG_RES_PATH} -h18_imeimd5 ${IMEIMD5_H_18_GUANGDIANTONG_RES_PATH} \
-dt_oneday_ago ${dt_oneday_ago}
......
......@@ -44,7 +44,7 @@ spark-submit --class mobvista.dmp.datasource.taobao.EtlOtherDataFromPostBackDail
--conf spark.sql.broadcastTimeout=1200 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 2g --executor-cores 6 --num-executors 150 \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 6 --num-executors 150 \
../../${JAR} \
-output ${OUTPUT_PATH} \
-dt_today ${dt_today} -dt_oneday_ago ${dt_oneday_ago}
......
......@@ -38,7 +38,7 @@ spark-submit --class mobvista.dmp.datasource.taobao.AlipayLaHuoDaily \
--conf spark.yarn.executor.memoryOverhead=4096 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 3g --executor-cores 6 --num-executors 120 ../../${JAR} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 6 --num-executors 120 ../../${JAR} \
-imeioutput "${ALIPAY_IMEIMD5_OUTPUT_PATH}" \
-today ${dt_today} -last_req_day ${last_req_day} -dt_after_one_day ${dt_after_one_day}\
-input_one_day ${INPUT_ONE_DAY} -input_two_day ${INPUT_TWO_DAY} -input_three_day ${INPUT_THREE_DAY} \
......
......@@ -34,7 +34,7 @@ spark-submit --class mobvista.dmp.datasource.taobao.AlipayTmpDataToDmp \
--conf spark.yarn.executor.memoryOverhead=4096 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 3g --executor-cores 4 --num-executors 40 \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 4 --num-executors 40 \
../../${JAR} -imeiRequestInput ${IMEIMD5_REQUEST_INPUT_PATH} -imeiResponseInput ${IMEIMD5_RESPONSE_INPUT_PATH} \
-output01 ${OUTPUT01} -output02 ${OUTPUT02}
......@@ -64,7 +64,7 @@ fi
# --conf spark.yarn.executor.memoryOverhead=4096 \
# --conf spark.sql.autoBroadcastJoinThreshold=31457280 \
# --files ${HIVE_SITE_PATH} \
# --master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 3g --executor-cores 4 --num-executors 40 \
# --master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 4 --num-executors 40 \
# ../../${JAR} -dt_today ${dt_today} -dt_three_days_ago ${dt_three_days_ago} \
# -ActivationOutput ${ACTIVATIONOUTPUT} -AcquisitionOutput ${ACQUISITIONOUTPUT}
#
......
......@@ -34,7 +34,7 @@ spark-submit --class mobvista.dmp.datasource.taobao.AlipayTmpDataToDmp \
--conf spark.yarn.executor.memoryOverhead=4096 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 3g --executor-cores 4 --num-executors 40 \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 4 --num-executors 40 \
../../${JAR} -imeiRequestInput ${IMEIMD5_REQUEST_INPUT_PATH} -imeiResponseInput ${IMEIMD5_RESPONSE_INPUT_PATH} \
-output01 ${OUTPUT01} -output02 ${OUTPUT02}
......@@ -64,7 +64,7 @@ fi
# --conf spark.yarn.executor.memoryOverhead=4096 \
# --conf spark.sql.autoBroadcastJoinThreshold=31457280 \
# --files ${HIVE_SITE_PATH} \
# --master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 3g --executor-cores 4 --num-executors 40 \
# --master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 4 --num-executors 40 \
# ../../${JAR} -dt_today ${dt_today} -dt_three_days_ago ${dt_three_days_ago} \
# -ActivationOutput ${ACTIVATIONOUTPUT} -AcquisitionOutput ${ACQUISITIONOUTPUT}
#
......
......@@ -34,7 +34,7 @@ spark-submit --class mobvista.dmp.datasource.taobao.AlipayTmpDataToDmp \
--conf spark.yarn.executor.memoryOverhead=4096 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 3g --executor-cores 4 --num-executors 40 \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 4 --num-executors 40 \
../../${JAR} -imeiRequestInput ${IMEIMD5_REQUEST_INPUT_PATH} -imeiResponseInput ${IMEIMD5_RESPONSE_INPUT_PATH} \
-output01 ${OUTPUT01} -output02 ${OUTPUT02}
......@@ -64,7 +64,7 @@ fi
# --conf spark.yarn.executor.memoryOverhead=4096 \
# --conf spark.sql.autoBroadcastJoinThreshold=31457280 \
# --files ${HIVE_SITE_PATH} \
# --master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 3g --executor-cores 4 --num-executors 40 \
# --master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 4 --num-executors 40 \
# ../../${JAR} -dt_today ${dt_today} -dt_three_days_ago ${dt_three_days_ago} \
# -ActivationOutput ${ACTIVATIONOUTPUT} -AcquisitionOutput ${ACQUISITIONOUTPUT}
#
......
......@@ -34,7 +34,7 @@ spark-submit --class mobvista.dmp.datasource.taobao.AlipayTmpDataToDmp \
--conf spark.yarn.executor.memoryOverhead=4096 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 3g --executor-cores 4 --num-executors 40 \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 4 --num-executors 40 \
../../${JAR} -imeiRequestInput ${IMEIMD5_REQUEST_INPUT_PATH} -imeiResponseInput ${IMEIMD5_RESPONSE_INPUT_PATH} \
-output01 ${OUTPUT01} -output02 ${OUTPUT02}
......@@ -64,7 +64,7 @@ fi
# --conf spark.yarn.executor.memoryOverhead=4096 \
# --conf spark.sql.autoBroadcastJoinThreshold=31457280 \
# --files ${HIVE_SITE_PATH} \
# --master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 3g --executor-cores 4 --num-executors 40 \
# --master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 4 --num-executors 40 \
# ../../${JAR} -dt_today ${dt_today} -dt_three_days_ago ${dt_three_days_ago} \
# -ActivationOutput ${ACTIVATIONOUTPUT} -AcquisitionOutput ${ACQUISITIONOUTPUT}
#
......
......@@ -30,7 +30,7 @@ spark-submit --class mobvista.dmp.datasource.taobao.AlipayOtherDataToDmp \
--conf spark.sql.broadcastTimeout=1200 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 2g --executor-cores 6 --num-executors 140 \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 6 --num-executors 140 \
../../${JAR} \
-output01 ${OUTPUT_PATH01} -output02 ${OUTPUT_PATH02} \
-dt_today ${dt_today} -dt_oneday_ago ${dt_oneday_ago} -hour ${hour}
......
......@@ -30,7 +30,7 @@ spark-submit --class mobvista.dmp.datasource.taobao.AlipayOtherDataToDmp \
--conf spark.sql.broadcastTimeout=1200 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 2g --executor-cores 6 --num-executors 140 \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 6 --num-executors 140 \
../../${JAR} \
-output01 ${OUTPUT_PATH01} -output02 ${OUTPUT_PATH02} \
-dt_today ${dt_today} -dt_oneday_ago ${dt_oneday_ago} -hour ${hour}
......
......@@ -30,7 +30,7 @@ spark-submit --class mobvista.dmp.datasource.taobao.AlipayOtherDataToDmp \
--conf spark.sql.broadcastTimeout=1200 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 2g --executor-cores 6 --num-executors 140 \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 6 --num-executors 140 \
../../${JAR} \
-output01 ${OUTPUT_PATH01} -output02 ${OUTPUT_PATH02} \
-dt_today ${dt_today} -dt_oneday_ago ${dt_oneday_ago} -hour ${hour}
......
......@@ -30,7 +30,7 @@ spark-submit --class mobvista.dmp.datasource.taobao.AlipayOtherDataToDmp \
--conf spark.sql.broadcastTimeout=1200 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 2g --executor-cores 6 --num-executors 140 \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 6 --num-executors 140 \
../../${JAR} \
-output01 ${OUTPUT_PATH01} -output02 ${OUTPUT_PATH02} \
-dt_today ${dt_today} -dt_oneday_ago ${dt_oneday_ago} -hour ${hour}
......
......@@ -37,7 +37,7 @@ spark-submit --class mobvista.dmp.datasource.taobao.EtlDealidDaily \
--conf spark.yarn.executor.memoryOverhead=4096 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 9g --driver-memory 3g --executor-cores 6 --num-executors 30 \
--master yarn --deploy-mode cluster --executor-memory 10g --driver-memory 4g --executor-cores 6 --num-executors 30 \
../${JAR} -dt_dash_today ${dt_dash_today} \
-oppooutput ${OPPO_OUTPUT} \
-inmobioutput ${INMOBI_OUTPUT}
......
......@@ -60,7 +60,7 @@ spark-submit --class mobvista.dmp.datasource.taobao.EtlLazadaActivitionDaily \
--conf spark.yarn.executor.memoryOverhead=4096 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 3g --executor-cores 4 --num-executors 90 ../${JAR} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 4 --num-executors 90 ../${JAR} \
-gaidoutput "${GAID_OUTPUT_PATH}" \
-today ${dt_today} \
-input_one_day ${INPUT_ONE_DAY} -input_two_day ${INPUT_TWO_DAY} -input_three_day ${INPUT_THREE_DAY} \
......@@ -105,7 +105,7 @@ fi
# --conf spark.yarn.executor.memoryOverhead=4096 \
# --conf spark.sql.autoBroadcastJoinThreshold=31457280 \
# --files ${HIVE_SITE_PATH} \
# --master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 3g --executor-cores 4 --num-executors 70 ../${JAR} \
# --master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 4 --num-executors 70 ../${JAR} \
# -gaidoutput "${GAID_OUTPUT_PATH}" -gaidinput "${GAID_INPUT_PATH}" -newoutput "${NEW_OUTPUT_PATH}" \
# -today ${dt_today} -dt_30days_ago ${dt_30days_ago}
#
......
......@@ -18,7 +18,7 @@ spark-submit --class mobvista.dmp.datasource.taobao.EtlComTencentNewsDaily \
--conf spark.sql.broadcastTimeout=1200 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 2g --executor-cores 6 --num-executors 120 \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 6g --executor-cores 6 --num-executors 120 \
../../${JAR} \
-output ${OUTPUT_PATH} -coalesce 500 \
-dt_today ${dt_today} -dt_dash_rec7day ${dt_dash_rec7day} -dt_dash_rec15day ${dt_dash_rec15day}
......
type=command
dependencies=uc_imei_lahuo_ck,uc_oaid_lahuo_request
dependencies=uc_imei_lahuo_ck
command=sh -x uc_imei_lahuo_request.sh
......@@ -25,7 +25,7 @@ spark-submit --class mobvista.dmp.datasource.taobao.UCLaHuoDaily \
--conf spark.yarn.executor.memoryOverhead=4096 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 3g --executor-cores 4 --num-executors 80 ../../${JAR} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 4 --num-executors 80 ../../${JAR} \
-imeioutput "${UC_IMEIMD5_OUTPUT_PATH}" -oaidoutput "${UC_OAIDMD5_OUTPUT_PATH}" \
-today ${dt_today} -last_req_day ${last_req_day}
......
......@@ -41,7 +41,7 @@ spark-submit --class mobvista.dmp.datasource.taobao.UCTmpDataToDMP \
--conf spark.yarn.executor.memoryOverhead=4096 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 3g --executor-cores 4 --num-executors 40 \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 4 --num-executors 40 \
../../${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} \
......
type=command
dependencies=uc_imei_lahuo_request
dependencies=uc_imei_lahuo_request,uc_oaid_lahuo_request
command=sh -x uc_lahuo_df.sh
......@@ -34,7 +34,7 @@ spark-submit --class mobvista.dmp.datasource.taobao.UCOtherDataToDmp \
--conf spark.sql.broadcastTimeout=1200 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 2g --executor-cores 6 --num-executors 150 \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 6 --num-executors 150 \
../../${JAR} \
-output01 ${OUTPUT_PATH01} -output02 ${OUTPUT_PATH02} \
-output03 ${OUTPUT_PATH03} -output04 ${OUTPUT_PATH04} \
......
......@@ -30,7 +30,7 @@ spark-submit --class mobvista.dmp.datasource.taobao.EtlH32FromDmInstallListV2 \
--conf spark.sql.broadcastTimeout=1200 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 2g --executor-cores 6 --num-executors 60 \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 6 --num-executors 60 \
../../${JAR} \
-h32_imei ${IMEI_H_32_GUANGDIANTONG_RES_PATH} -h32_imeimd5 ${IMEIMD5_H_32_GUANGDIANTONG_RES_PATH} \
-dt_oneday_ago ${dt_oneday_ago}
......
......@@ -36,7 +36,7 @@ spark-submit --class mobvista.dmp.datasource.taobao.YOUKULaXinDaily \
--conf spark.yarn.executor.memoryOverhead=4096 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 3g --executor-cores 4 --num-executors 80 ../../${JAR} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 4 --num-executors 80 ../../${JAR} \
-imeioutput "${YOUKU_IMEIMD5_OUTPUT_PATH}" -oaidoutput "${YOUKU_OAIDMD5_OUTPUT_PATH}" \
-input_one_day ${INPUT_ONE_DAY} -input_two_day ${INPUT_TWO_DAY} -input_three_day ${INPUT_THREE_DAY} \
-oaid_input_one_day ${OAID_INPUT_ONE_DAY} -oaid_input_two_day ${OAID_INPUT_TWO_DAY} -oaid_input_three_day ${OAID_INPUT_THREE_DAY} \
......
......@@ -26,7 +26,7 @@ spark-submit --class mobvista.dmp.datasource.taobao.YoukuTmpDataToDmp \
--conf spark.yarn.executor.memoryOverhead=4096 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 3g --executor-cores 4 --num-executors 40 \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 4 --num-executors 40 \
../../${JAR} -Input ${INPUT_PATH} -Output ${OUTPUT_PATH} \
-update ${update}
......@@ -54,7 +54,7 @@ spark-submit --class mobvista.dmp.datasource.taobao.YoukuLaXinPollingDataDedupli
--conf spark.yarn.executor.memoryOverhead=4096 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 3g --executor-cores 4 --num-executors 40 \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 4 --num-executors 40 \
../../${JAR} -dt_today ${dt_today} -dt_begin_days ${dt_begin_days} \
-AcquisitionOutput ${ACQUISITIONOUTPUT}
......
......@@ -21,7 +21,7 @@ EXPIRE_PATH="${PACKAGE_BLACK_LIST}/$expire_path"
spark-submit --class mobvista.dmp.main.PackageBlackList \
--conf spark.sql.shuffle.partitions=10 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 4g --driver-memory 2g --executor-cores 2 --num-executors 5 \
--master yarn --deploy-mode cluster --executor-memory 4g --driver-memory 4g --executor-cores 2 --num-executors 5 \
../${JAR} \
-date "${yes_date}" \
-iosDailyPath "${TMP_IOS_APP_INFO_PATH}" -adrDailyPath "${TMP_ADR_APP_INFO_PATH}" \
......
......@@ -64,7 +64,7 @@ $HIVE_CMD -v -hivevar dt_today ${dt_today} -hivevar update_date ${dt_today
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--jars s3://mob-emr-test/dataplatform/DataWareHouse/offline/myjar/hive-hcatalog-core-2.3.3.jar \
--master yarn --deploy-mode cluster --name apps_flyer_total --executor-memory 4g --driver-memory 2g --executor-cores 3 --num-executors 5 \
--master yarn --deploy-mode cluster --name apps_flyer_total --executor-memory 4g --driver-memory 4g --executor-cores 3 --num-executors 5 \
../${JAR} -outputtotal ${OUTPUT_TOTAL_PATH} -dmpuserinfo ${DMP_USER_INFO_OUTPUT_PATH} \
-coalesce 20 \
-today ${dt_today} -update_date ${dt_today_dash}
......
......@@ -23,7 +23,7 @@ if [ $? -eq 0 ];then
hadoop fs -rm -r $OUTPUT_PATH
# find unmanned
spark-submit --master yarn --deploy-mode cluster --executor-memory 4g --driver-memory 3g --executor-cores 2 --num-executors 15 \
spark-submit --master yarn --deploy-mode cluster --executor-memory 4g --driver-memory 4g --executor-cores 2 --num-executors 15 \
--conf spark.yarn.executor.memoryOverhead=2048M \
--class mobvista.dmp.main.FindUnmatchBundle ../${JAR} -input $INPUT_PATH -output $OUTPUT_PATH
if [ $? -ne 0 ];then
......
......@@ -24,7 +24,7 @@ hadoop fs -rm -r "$OUTPUT_PATH/"
spark-submit --class mobvista.dmp.datasource.clever.ParseCleverDaily \
--conf spark.yarn.executor.memoryOverhead=2048 --conf spark.network.timeout=720s \
--master yarn --deploy-mode cluster --executor-memory 4g --driver-memory 3g --executor-cores 2 --num-executors 40 \
--master yarn --deploy-mode cluster --executor-memory 4g --driver-memory 4g --executor-cores 2 --num-executors 40 \
../${JAR} -input $INPUT_PATH -output $OUTPUT_PATH -parallelism 100 -coalesce 20
if [ $? -ne 0 ];then
exit 255
......
......@@ -26,7 +26,7 @@ hadoop fs -rm -r "$OUTPUT_PATH/"
spark-submit --class mobvista.dmp.datasource.clever.CleverInstallList \
--conf spark.yarn.executor.memoryOverhead=2048 --conf spark.network.timeout=720s --conf spark.app.tag=-1 \
--master yarn --deploy-mode cluster --executor-memory 5g --driver-memory 3g --executor-cores 2 --num-executors 20 \
--master yarn --deploy-mode cluster --executor-memory 6g --driver-memory 4g --executor-cores 2 --num-executors 20 \
../${JAR} -input $INPUT_PATH -oldInput $OLD_INPUT_PATH -output $OUTPUT_PATH -date $dt -parallelism 200 -coalesce 20
if [ $? -ne 0 ];then
exit 255
......
......@@ -27,7 +27,7 @@ spark-submit --class mobvista.dmp.datasource.dm.AgePackageNames \
--conf spark.sql.shuffle.partitions=3000 \
--conf spark.network.timeout=720s \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 5g --driver-memory 5g --executor-cores 2 --num-executors 220 \
--master yarn --deploy-mode cluster --executor-memory 6g --driver-memory 4g --executor-cores 2 --num-executors 220 \
../../${JAR} -dt_today ${dt_today} -update ${update} -Age_Package_Names ${Age_Package_Names} -output01 ${OUTPUT_PATH01} -output02 ${OUTPUT_PATH02} -coalesce 200
......
......@@ -27,7 +27,7 @@ spark-submit --class mobvista.dmp.datasource.dm.CanglanPackageNames \
--conf spark.sql.shuffle.partitions=3000 \
--conf spark.network.timeout=720s \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 5g --driver-memory 5g --executor-cores 2 --num-executors 220 \
--master yarn --deploy-mode cluster --executor-memory 6g --driver-memory 4g --executor-cores 2 --num-executors 220 \
../../${JAR} -dt_today ${dt_today} -update ${update} -Canglan_Package_Names ${Canglan_Package_Names} -output01 ${OUTPUT_PATH01} -output02 ${OUTPUT_PATH02} -coalesce 200
......
......@@ -27,7 +27,7 @@ spark-submit --class mobvista.dmp.datasource.dm.ThreeKingdomsGame \
--conf spark.sql.shuffle.partitions=3000 \
--conf spark.network.timeout=720s \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 5g --driver-memory 5g --executor-cores 2 --num-executors 220 \
--master yarn --deploy-mode cluster --executor-memory 6g --driver-memory 4g --executor-cores 2 --num-executors 220 \
../../${JAR} -dt_today ${dt_today} -update ${update} -package_names_input ${Three_Kingdoms_Package_Names} -output01 ${OUTPUT_PATH01} -output02 ${OUTPUT_PATH02} -coalesce 200
......
......@@ -28,7 +28,7 @@ spark-submit --class mobvista.dmp.datasource.dm.ShinnyPackageNames \
--conf spark.sql.shuffle.partitions=3000 \
--conf spark.network.timeout=720s \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 5g --driver-memory 5g --executor-cores 2 --num-executors 220 \
--master yarn --deploy-mode cluster --executor-memory 6g --driver-memory 4g --executor-cores 2 --num-executors 220 \
../../${JAR} -dt_today ${dt_today} -update ${update} -Shinny_Package_Names ${Shinny_Package_Names} -output01 ${OUTPUT_PATH01} -output02 ${OUTPUT_PATH02} -coalesce 200
......
......@@ -825,7 +825,7 @@ userInfoJob() {
--conf spark.speculation.multiplier=1 \
--jars ${JARS} \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 5g --driver-memory 3g --executor-cores 2 --num-executors 20 \
--master yarn --deploy-mode cluster --executor-memory 6g --driver-memory 4g --executor-cores 2 --num-executors 20 \
${jar} -date $LOG_TIME -dailyPath $dailyPath -agePath $agePath -genderPath $genderPath \
-dailyFormat ${dailyFormat} -dailyDidIndex $dailyDidIndex -dailyDidTypeIndex $dailyDidTypeIndex -dailyPltIndex $dailyPltIndex -dailyCountryIndex $dailyCountryIndex \
-outputPath $outputPath -parallelism ${parallelism} -coalesce ${coalesce}
......@@ -874,7 +874,7 @@ userInfoJob_dsp_req() {
--conf spark.speculation.multiplier=1 \
--jars ${JARS} \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 10g --driver-memory 3g --executor-cores 4 --num-executors 80 \
--master yarn --deploy-mode cluster --executor-memory 10g --driver-memory 4g --executor-cores 4 --num-executors 80 \
${jar} -date $LOG_TIME -dailyPath $dailyPath -agePath $agePath -genderPath $genderPath \
-dailyFormat ${dailyFormat} -dailyDidIndex $dailyDidIndex -dailyDidTypeIndex $dailyDidTypeIndex -dailyPltIndex $dailyPltIndex -dailyCountryIndex $dailyCountryIndex \
-outputPath $outputPath -parallelism ${parallelism} -coalesce ${coalesce}
......
......@@ -22,7 +22,7 @@ hadoop fs -rm -r $OUTPUT_PATH
spark-submit --class mobvista.dmp.datasource.address.AddressInfoTotal \
--conf spark.yarn.executor.memoryOverhead=1024 --conf spark.network.timeout=720s \
--conf spark.sql.shuffle.partitions=300 \
--master yarn --deploy-mode cluster --executor-memory 4g --driver-memory 3g --executor-cores 2 --num-executors 100 \
--master yarn --deploy-mode cluster --executor-memory 4g --driver-memory 4g --executor-cores 2 --num-executors 100 \
../${JAR} -input ${INPUT_PATH} -output ${OUTPUT_PATH} -dailyFormat "text" -parallelism 200 -coalesce 20 \
-indices "0,2,3,4,5,6,7"
......
......@@ -31,7 +31,7 @@ spark-submit --class mobvista.dmp.datasource.behavior.ThirdPartySourceDaily \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--jars s3://mob-emr-test/dataplatform/DataWareHouse/offline/myjar/hive-hcatalog-core-2.3.3.jar \
--master yarn --deploy-mode cluster --name behavior_from_third_party_daily --executor-memory 2g --driver-memory 2g --executor-cores 2 --num-executors 2 \
--master yarn --deploy-mode cluster --name behavior_from_third_party_daily --executor-memory 2g --driver-memory 4g --executor-cores 2 --num-executors 2 \
../${JAR} -outputtotal ${OUTPUT_TOTAL_PATH} -coalesce 10 \
-today ${dt_today} -yesbef3 ${dt_yes_bef3}
......
......@@ -30,7 +30,7 @@ spark-submit --class mobvista.dmp.datasource.behavior.ThirdPartySourceTotal \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--jars s3://mob-emr-test/dataplatform/DataWareHouse/offline/myjar/hive-hcatalog-core-2.3.3.jar \
--master yarn --deploy-mode cluster --name behavior_from_third_party_total --executor-memory 2g --driver-memory 2g --executor-cores 2 --num-executors 2 \
--master yarn --deploy-mode cluster --name behavior_from_third_party_total --executor-memory 2g --driver-memory 4g --executor-cores 2 --num-executors 2 \
../${JAR} -outputtotal ${OUTPUT_TOTAL_PATH} -dmpevent ${DMP_EVENT_TAG_PATH} -coalesce 10 \
-yesterday ${dt_yesterday}
......
......@@ -37,7 +37,7 @@ if [[ $? -ne 0 ]]; then
exit 255
fi
sleep $((fors * 20))
sleep $((fors * 25))
shell=" -cp /root/workspace/DMP-1.0.3-jar-with-dependencies.jar mobvista.dmp.datasource.iqiyi.IQiYiRequest"
......
......@@ -28,7 +28,7 @@ spark-submit --class mobvista.dmp.datasource.iqiyi.IQiYiTmpDataToDMP \
--conf spark.yarn.executor.memoryOverhead=4096 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 3g --executor-cores 4 --num-executors 30 \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 4 --num-executors 30 \
../${JAR} -input ${INPUT} \
-output ${OUTPUT} \
-update ${update}
......@@ -53,7 +53,7 @@ spark-submit --class mobvista.dmp.datasource.iqiyi.IQiYiLaHuoFourDaysDataDedupli
--conf spark.yarn.executor.memoryOverhead=4096 \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 3g --executor-cores 4 --num-executors 40 \
--master yarn --deploy-mode cluster --executor-memory 8g --driver-memory 4g --executor-cores 4 --num-executors 40 \
../${JAR} -dt_today ${dt_today} -dt_three_days_ago ${dt_three_days_ago} \
-output ${FOUR_DAYS_OUTPUT}
......
......@@ -39,7 +39,7 @@ spark-submit --class mobvista.dmp.datasource.setting.SettingTotal \
--conf spark.sql.autoBroadcastJoinThreshold=31457280 \
--files ${HIVE_SITE_PATH} \
--jars /data/hadoop-alternative/hive/auxlib/Common-SerDe-1.0-SNAPSHOT.jar \
--master yarn --deploy-mode cluster --name apps_flyer_total --executor-memory 4g --driver-memory 2g --executor-cores 3 --num-executors 5 \
--master yarn --deploy-mode cluster --name apps_flyer_total --executor-memory 4g --driver-memory 4g --executor-cores 3 --num-executors 5 \
../${JAR} -outputtotal ${APP_ID_MAPPING_TMP} \
-coalesce 30 \
-today ${LOG_TIME}
......
......@@ -28,7 +28,7 @@ installPath="s3://mob-emr-test/dataplatform/DataWareHouse/data/dwh/dm_install_li
outputPath="s3://mob-emr-test/feng.liang/report/activeInstall"
spark-submit --master yarn --executor-cores 2 --executor-memory 4g \
--class mobvista.prd.main.ActiveInstall --driver-memory 3g ../${JAR} \
--class mobvista.prd.main.ActiveInstall --driver-memory 4g ../${JAR} \
$date $appTagPath $installPath $outputPath
if [ $? -ne 0 ];then
......
......@@ -12,7 +12,7 @@ while [ $startDay -le $stopDay ];do
outputPath="s3://mob-emr-test/feng.liang/weightGame/$date_path"
hadoop fs -rm -r $outputPath
spark-submit --class mobvista.prd.main.WeightGame \
--master yarn --executor-memory 6g --driver-memory 3g --executor-cores 2 --num-executors 50 \
--master yarn --executor-memory 6g --driver-memory 4g --executor-cores 2 --num-executors 50 \
../DMP.jar "$inputPath" "$installPath" "$outputPath"
if [ $? -ne 0 ];then
echo "$startDay fail"
......
Markdown is supported
0% or
You are about to add 0 people to the discussion. Proceed with caution.
Finish editing this message first!
Please register or to comment