fix rtdmp

azkaban/rtdmp/rtdmp.sh

@@ -4,48 +4,35 @@ source ../dmp_env.sh
 
 today=${ScheduleTime}
 
-date_time=$(date +"%Y-%m-%d.%H" -d "-1 hour $today")
+date_time=$(date +"%Y%m%d%H" -d "-2 hour $today")
 
 date_path=$(date +%Y/%m/%d/%H -d "-1 hour $today")
 
-part_num=$(hadoop fs -ls s3://mob-emr-test/dataplatform/rtdmp_pre/${date_path}/ | wc -l)
-if [[ ${part_num} -le 50 ]]; then
-  echo "This Dir No Data !!!"
-  partition=10
-  coalesce=10
-  executor=2
-  memory=4
-  core=2
-  flag=0
-else
-  partition=2000
-  coalesce=200
-  executor=8
-  memory=10
-  core=4
-  flag=1
-fi
 INPUT="s3://mob-emr-test/dataplatform/rtdmp_pre/${date_path}"
 
-OUTPUT="s3://mob-emr-test/dataplatform/rtdmp_deal/${date_path}"
-
-before_date_path=$(date +%Y/%m/%d/%H -d "-2 hour $today")
-
-BEFORE_OUTPUT="s3://mob-emr-test/dataplatform/rtdmp/${before_date_path}"
-
-check_await "${BEFORE_OUTPUT}/_SUCCESS"
+OUTPUT="s3://mob-emr-test/dataplatform/DataWareHouse/data/dwh/audience_merge/${date_path}"
 
 spark-submit --class mobvista.dmp.datasource.rtdmp.RTDmpMain \
   --name "RTDmpMain.${date_time}" \
-  --conf spark.sql.shuffle.partitions=${partition} \
-  --conf spark.default.parallelism=${partition} \
+  --conf spark.sql.shuffle.partitions=1000 \
+  --conf spark.default.parallelism=1000 \
   --conf spark.kryoserializer.buffer.max=512m \
   --conf spark.kryoserializer.buffer=64m \
   --master yarn --deploy-mode cluster \
-  --executor-memory ${memory}g --driver-memory 6g --executor-cores ${core} --num-executors ${executor} \
+  --executor-memory 18g --driver-memory 4g --executor-cores 5 --num-executors 40 \
   .././DMP.jar \
-  -flag ${flag} -time ${date_time} -input ${INPUT} -output ${OUTPUT} -coalesce ${coalesce}
+  -datetime ${date_time} -input ${INPUT} -output ${OUTPUT} -coalesce 200
 
 if [[ $? -ne 0 ]]; then
   exit 255
 fi
+
+mount_partition "audience_merge" "dt='${curr_time}'" "$OUTPUT"
+
+expire_time=$(date +"%Y%m%d%H" -d "-24 hour $today")
+
+expire_date_path=$(date +%Y/%m/%d/%H -d "-24 hour $today")
+
+EXPIRE_OUTPUT_PATH="s3://mob-emr-test/dataplatform/DataWareHouse/data/dwh/audience_merge/${expire_date_path}"
+
+unmount_partition "audience_merge" "dt='${expire_time}'" "${EXPIRE_OUTPUT_PATH}"
\ No newline at end of file

azkaban/rtdmp/rtdmp_fetch.sh

@@ -4,7 +4,7 @@ source ../dmp_env.sh
 
 today=${ScheduleTime:-$1}
 
-start_time=$(date +"%Y-%m-%d %H:00:00" -d "-24 hours $today")
+start_time=$(date +"%Y-%m-%d %H:00:00" -d "-168 hours $today")
 end_time=$(date +"%Y-%m-%d %H:59:59" -d "-1 hours $today")
 
 java -cp ../${JAR} mobvista.dmp.datasource.rtdmp.ServerMain "${start_time}" "${end_time}"

azkaban/rtdmp/rtdmp_merge.job

 type=command
-command=bash -x rtdmp_merge.sh
+dependencies=rtdmp_pre
\ No newline at end of file
+command=bash -x rtdmp.sh
\ No newline at end of file

azkaban/rtdmp/rtdmp_merge_ck.sh

@@ -12,19 +12,17 @@ database="dwh"
 table="audience_merge"
 
 spark-submit --class mobvista.dmp.datasource.rtdmp.RTDmpMergeCK \
-     --name "RTDmpMergeCK.wangjf.${date_time}" \
-     --conf spark.sql.shuffle.partitions=1000 \
-     --conf spark.default.parallelism=1000 \
-     --conf spark.kryoserializer.buffer.max=256m \
-     --conf spark.speculation=false \
-     --conf spark.speculation.quantile=0.9 \
-     --conf spark.speculation.multiplier=1.3 \
-     --conf spark.executor.extraJavaOptions="-XX:+UseG1GC" \
-     --files ${HIVE_SITE_PATH} \
-     --jars ${SPARK_HOME}/auxlib/Common-SerDe-1.0-SNAPSHOT.jar \
-     --master yarn --deploy-mode cluster --executor-memory 10g --driver-memory 6g  --executor-cores 5 --num-executors 6 \
-     ../${JAR} -date_time "${date_time}" -host ${host} -cluster ${cluster} -database ${database} -table ${table}
+  --name "RTDmpMergeCK.wangjf.${date_time}" \
+  --conf spark.sql.shuffle.partitions=1000 \
+  --conf spark.default.parallelism=1000 \
+  --conf spark.kryoserializer.buffer.max=256m \
+  --conf spark.speculation=false \
+  --conf spark.speculation.quantile=0.9 \
+  --conf spark.speculation.multiplier=1.3 \
+  --conf spark.executor.extraJavaOptions="-XX:+UseG1GC" \
+  --master yarn --deploy-mode cluster --executor-memory 10g --driver-memory 6g --executor-cores 6 --num-executors 10 \
+  ../${JAR} -date_time "${date_time}" -host ${host} -cluster ${cluster} -database ${database} -table ${table}
 
 if [[ $? -ne 0 ]]; then
-    exit 255
-fi
+  exit 255
+fi
\ No newline at end of file