[
https://issues.apache.org/jira/browse/SPARK-25367?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]
yy updated SPARK-25367:
-----------------------
Description:
We save the dataframe object as a hive table in orc/parquet format in the spark
shell.
After we modified the column type (int to double) of this table in hive jdbc,
we found the column type queried in spark-shell didn't change, but changed in
hive jdbc. After we restarted the spark-shell, this table's column type is
still incompatible as showed in hive jdbc.
The coding process are as follows:
spark-shell:
{code:java}
val df = spark.read.json("examples/src/main/resources/people.json");
df.write.format("orc").saveAsTable("people_test");
spark.catalog.refreshTable("people_test")
spark.sql("desc people_test").show()
+--------+---------+-------+
|col_name|data_type|comment|
+--------+---------+-------+
| age| bigint| null|
| name| string| null|
+--------+---------+-------+
{code}
hive:
{code:java}
hive> desc people_test;
OK
age bigint
name string
Time taken: 0.454 seconds, Fetched: 2 row(s)
hive> alter table people_test change column age age1 double;
OK
Time taken: 0.68 seconds
hive> desc people_test;
OK
age1 double
name string
Time taken: 0.358 seconds, Fetched: 2 row(s){code}
spark-shell:
{code:java}
spark.sql("desc people_test").show()
+--------+---------+-------+
|col_name|data_type|comment|
+--------+---------+-------+
| age| bigint| null|
| name| string| null|
+--------+---------+-------+
{code}
We also tested in spark-shell by creating a table using spark.sql("create table
XXX()"), the modified columns are consistent.
was:
We save the dataframe object as a hive table in orc/parquet format in the spark
shell.
After we modified the column type (int to double) of this table in hive jdbc,
we found the column type queried in spark-shell didn't change, but changed in
hive jdbc. After we restarted the spark-shell, this table's column type is
still incompatible as showed in hive jdbc.
The coding process are as follows:
spark-shell:
{code:java}
val df = spark.read.json("examples/src/main/resources/people.json");
df.write.format("orc").saveAsTable("people_test");
spark.catalog.refreshTable("people_test")
spark.sql("desc people").show()
{code}
hive:
{code:java}
alter table people_test change column age age1 double;
desc people_test;{code}
spark-shell:
{code:java}
spark.sql("desc people").show()
{code}
We also tested in spark-shell by creating a table using spark.sql("create table
XXX()"), the modified columns are consistent.
> Hive table created by Spark dataFrame has incompatiable schema in spark and
> hive
> --------------------------------------------------------------------------------
>
> Key: SPARK-25367
> URL: https://issues.apache.org/jira/browse/SPARK-25367
> Project: Spark
> Issue Type: Bug
> Components: Spark Shell, SQL
> Affects Versions: 2.2.0, 2.2.1, 2.2.2, 2.3.0, 2.3.1
> Environment: spark2.2.1-hadoop-2.6.0-chd-5.4.2
> hive-1.2.1
> Reporter: yy
> Priority: Major
> Labels: sparksql
> Fix For: 2.3.2
>
>
> We save the dataframe object as a hive table in orc/parquet format in the
> spark shell.
> After we modified the column type (int to double) of this table in hive
> jdbc, we found the column type queried in spark-shell didn't change, but
> changed in hive jdbc. After we restarted the spark-shell, this table's column
> type is still incompatible as showed in hive jdbc.
> The coding process are as follows:
> spark-shell:
> {code:java}
> val df = spark.read.json("examples/src/main/resources/people.json");
> df.write.format("orc").saveAsTable("people_test");
> spark.catalog.refreshTable("people_test")
> spark.sql("desc people_test").show()
> +--------+---------+-------+
> |col_name|data_type|comment|
> +--------+---------+-------+
> | age| bigint| null|
> | name| string| null|
> +--------+---------+-------+
> {code}
>
> hive:
> {code:java}
> hive> desc people_test;
> OK
> age bigint
> name string
> Time taken: 0.454 seconds, Fetched: 2 row(s)
> hive> alter table people_test change column age age1 double;
> OK
> Time taken: 0.68 seconds
> hive> desc people_test;
> OK
> age1 double
> name string
> Time taken: 0.358 seconds, Fetched: 2 row(s){code}
> spark-shell:
> {code:java}
> spark.sql("desc people_test").show()
> +--------+---------+-------+
> |col_name|data_type|comment|
> +--------+---------+-------+
> | age| bigint| null|
> | name| string| null|
> +--------+---------+-------+
> {code}
>
> We also tested in spark-shell by creating a table using spark.sql("create
> table XXX()"), the modified columns are consistent.
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
---------------------------------------------------------------------
To unsubscribe, e-mail: issues-unsubscr...@spark.apache.org
For additional commands, e-mail: issues-h...@spark.apache.org
