Hive插入从Hive CLI可以正常工作，但在终端上失败

Question

I currently have the following situation:

I have a shellscript that creates two tables and then fills one of them with data of the other.

my script looks somewhat like this:

    hive -e "CREATE EXTERNAL TABLE table1 ... ROW FORMAT DELIMITED FIELDS TERMINATED BY '\t' LOCATION '/user/input/'"
    hive -e "CREATE EXTERNAL TABLE table2 ... PARTITIONED BY ..."
    hive -e "WITH data AS (SELECT date, ...) FROM data INSERT OVERWRITE TABLE table2 PARTITION(part_date) SELECT ... date"

and then some more shellscript that selects certain data from table2. I use shellscript because there is some logic that has to be applied before I can do the selects on table2 .

The script runs without error, the tables are created, table1 has data in it but table2 ends up empty. For some reason it works when I use a very small test dataset, but as soon as the dataset becomes bigger (>1GB) table2 is empty.

If I run the very same commands from Hives CLI manually everything works fine and table2 has the expected data in it.

Why does this happen, and how could i resolve this?

Answer 1

Your shell script commands must be executed sequentially to get data in table2 .

Try this in your shell script:

hive -e "your first query" && 
hive -e "your second query" &&
hive -e "your third query"

This should execute your hive queries one after another. (2nd query waits for 1st to finish and 3rd query waits for both 1st and 2nd to finish)