Browse Source

feat:添加删除数据的脚本

zhaohaipeng 1 month ago
parent
commit
0000feff13
1 changed files with 16 additions and 32 deletions
  1. 16 32
      ad/30_delete_timer_file.sh

+ 16 - 32
ad/30_delete_timer_file.sh

@@ -4,32 +4,12 @@ export HADOOP_CONF_DIR=/etc/taihao-apps/hadoop-conf
 
 HADOOP=/opt/apps/HADOOP-COMMON/hadoop-common-current/bin/hadoop
 
-PREDICT_HOME=/root/zhaohp/recommend-emr-dataprocess/predict
-origin_data_hdfs_dir=/dw/recommend/model/31_ad_sample_data_v3_auto
-bucket_feature_hdfs_dir=/dw/recommend/model/33_ad_train_data_v3_auto
+origin_data_hdfs_dir=/dw/recommend/model/41_recsys_origin_date
+str_train_data_dir=/dw/recommend/model/41_recsys_str_train_data
+ros_train_data_dir=/dw/recommend/model/41_recsys_ros_train_data
 
-
-# 删除五天之前的预测结果文件
-delete_predict_5d_ago() {
-
-    echo "=========== 开始删除五天前的预测结果文件 $(date "+%Y-%m-%d %H:%M:%d") ==========="
-
-    tmp_file_name=./files_to_delete.txt
-
-    # 查询五天前的预测结果文件,并保存到临时文件
-    find "$PREDICT_HOME" -type f -mtime +5 > "${tmp_file_name}"
-
-    # 逐行读取临时文件中的路径并删除文件
-    while IFS= read -r file; do
-        echo "Deleting: $file"
-        rm -f "$file"
-    done < "${tmp_file_name}"
-
-    # 删除临时文件
-    rm -f "${tmp_file_name}"
-
-    echo "=========== 删除五天前的预测结果文件结束 $(date "+%Y-%m-%d %H:%M:%d") ==========="
-}
+str_bucket_data_dir=/dw/recommend/model/43_recsys_str_data_bucket
+ros_bucket_data_dir=/dw/recommend/model/43_recsys_ros_data_bucket
 
 # 删除HDFS中的目录
 delete_hdfs_path() {
@@ -52,7 +32,7 @@ delete_hdfs_path() {
         echo "${line}"
         if [[ "${modified_date}" < "${EARLY_DAYS_AGO}" ]]; then
             echo "Deleting: ${dir}"
-            $HADOOP fs -rm -r -skipTrash ${dir}
+#            $HADOOP fs -rm -r -skipTrash ${dir}
         fi
 
     done
@@ -63,12 +43,16 @@ delete_hdfs_path() {
 
 
 main() {
-    # 删除五天前的预测结果文件
-    delete_predict_5d_ago
-    # 删除七天之前的HDFS中的特征原始数据
-    delete_hdfs_path 7 $origin_data_hdfs_dir
-    # 删除七天之前的HDFS中的特征分桶数据
-    delete_hdfs_path 7 $bucket_feature_hdfs_dir
+    # 删除推荐原始数据
+    delete_hdfs_path 10 $origin_data_hdfs_dir
+
+    # 删除STR模型数据
+    delete_hdfs_path 10 $str_train_data_dir
+    delete_hdfs_path 10 $str_bucket_data_dir
+
+    # 删除ROS模型数据
+    delete_hdfs_path 10 $ros_train_data_dir
+    delete_hdfs_path 10 $ros_bucket_data_dir
 }