maropu commented on a change in pull request #28953:
URL: https://github.com/apache/spark/pull/28953#discussion_r451339072
##########
File path: docs/sql-data-sources-jdbc.md
##########
@@ -156,6 +156,20 @@ the following case-insensitive options:
</td>
</tr>
+ <tr>
+ <td><code>preActions</code></td>
+ <td>
+ Custom queries which you want to run before reading data from JDBC or
writing data to JDBC. Only DDL or DML (insert/update/delete) are allowed. It is
called per DataFrame, not per session. You can specify multiple queries
separated by semicolon. When exceptions occur in preActions, the queries in
preActions will be rollbacked.
Review comment:
>> It is called per DataFrame, not per session.
I think this statement looks ambiguous. What does `per DataFrame` mean?
Probably, `per session` refers to `sessionInitStatement` though. `postAction`
seems to be executed during the resolution phase in read paths, so...
```
$ ./bin/spark-shell
scala> paste:
sql("""
CREATE TABLE jdbcTable
USING org.apache.spark.sql.jdbc
OPTIONS (
driver "org.postgresql.Driver",
url "jdbc:postgresql:postgres",
dbtable "t1",
user 'maropu',
password ''
)
""")
<-- `postAction` executed
:quit
// Re-launch spark-shell
$ ./bin/spark-shell
scala> sql("select * from jdbcTable").show()
<-- `postAction` executed
scala> sql("select * from jdbcTable").show()
<-- `postAction` not executed
```
Is this an expected behaviour? On the other hand, in write paths, are
`preAction`/`postAction` executed per session? I'm afraid that users will get
confused about the current execution timing.
One more question; is there any usecase `preActions` can handle and
`sessionInitStatement` cannot?
----------------------------------------------------------------
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.
For queries about this service, please contact Infrastructure at:
[email protected]
---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]
