E-MapReduce:Sqoop FAQ

Timestamp shifts 8 hours after import

• Problem

  1. For example, the test_table table in an ApsaraDB RDS data source contains a timestamp field. When you query the source table test_table in DMS, the result contains five records with the id and applied_at columns. The values in the applied_at column range from 2024-06-14 10:08:04 to 2024-06-18 10:08:04 (incremented by one day, with the time fixed at 10:08:04).

  2. Run the following command to import data from test_table to HDFS.

     sqoop import \
     --connect jdbc:mysql://rm-2ze****341.mysql.rds.aliyuncs.com:3306/s***o_sqoop_db \
     --username s***o \
     --password ****** \
     --table test_table \
     --target-dir /user/hadoop/output \
     --delete-target-dir \
     --direct \
     --split-by id \
     --fields-terminated-by '|' \
     -m 1

  3. Query the import result.

     The result shows that the timestamp field in HDFS is shifted back by 8 hours.

     [root@master-1-1(xxx) ~]# hdfs dfs -cat /user/hadoop/output/part-m-00000
     1	2024-06-18 02:08:04
     2	2024-06-17 02:08:04
     3	2024-06-16 02:08:04
     4	2024-06-15 02:08:04
     5	2024-06-14 02:08:04

• Solution: When importing data that contains a timestamp field to HDFS, remove the --direct argument.

  sqoop import \
  --connect jdbc:mysql://rm-2ze****341.mysql.rds.aliyuncs.com:3306/s***o_sqoop_db \
  --username s***o \
  --password ****** \
  --table test_table \
  --target-dir /user/hadoop/output \
  --delete-target-dir \
  --split-by id \
  --fields-terminated-by '|' \
  -m 1

  The query result now shows the correct values.

  [root@master-1-1(xxx) ~]# hdfs dfs -cat /user/hadoop/output1/part-m-00000
  1	2024-06-18 10:08:04.0
  2	2024-06-17 10:08:04.0
  3	2024-06-16 10:08:04.0
  4	2024-06-15 10:08:04.0
  5	2024-06-14 10:08:04.0