Add JDBC Interpreter for Mysql

[1teed]

I added jdbc interpreter for mysql

I used Abstract Factory Pattern
It has the advantage that can be separated using the JDBC Connection only argument value

How to add your jdbc driver

1. add dependency for your jdbc driver (pom.xml)
2. implement Sub-Class (like MysqlConnection)
3. add branch connection in Factory Class (DBConnectionFactory)

It can be extended any jdbc. Such as Oracle, HIVE, Tajo, hBase, MariaDB, ...

Set interpreter

How to Use

[Leemoonsoo]

Thanks for the contribution.

For default value of zeppelin-site.xml, you'll need add JDBCInterpreter to zeppelin-zengine/src/main/java/org/apache/conf/ZeppelinConfiguration.java L384, ZEPPELIN_INTERPRETERS.

And about mysql dependency in jdbc/pom.xml file,

 <dependency>
<groupId>mysql</groupId>
<artifactId>mysql-connector-java</artifactId>
<version>5.1.6</version>
</dependency>

I think the dependency's version is The GNU General Public License, Version 2 and should not be used.

[1teed]

@Leemoonsoo

Thanks for your comment.
I fix mysql dependency to meet The GNU General Public License, Version 2 in pom.xml.
And add JDBCInterpreter to ZEPPELIN_INTERPRETERS in zeppelin-zengine/src/main/java/org/apache/conf/ZeppelinConfiguration.java.

[sylvinho81]

Hi @1teed,

I am trying to apply this patch and test it, because I think is really interesting and I need it.
After apply the changes and run:
mvn clean package
mvn install -DskipTests -Dspark.version=1.1.1 -Dhadoop.version=1.2.1

I can see:

[INFO] ------------------------------------------------------------------------
[INFO] Reactor Summary:
[INFO]
[INFO] Zeppelin ........................................... SUCCESS [ 3.472 s]
[INFO] Zeppelin: Interpreter .............................. SUCCESS [ 9.269 s]
[INFO] Zeppelin: Zengine .................................. SUCCESS [ 3.835 s]
[INFO] Zeppelin: Spark .................................... SUCCESS [ 23.308 s]
[INFO] Zeppelin: Markdown interpreter ..................... SUCCESS [ 2.025 s]
[INFO] Zeppelin: Angular interpreter ...................... SUCCESS [ 1.958 s]
[INFO] Zeppelin: Shell interpreter ........................ SUCCESS [ 1.993 s]
[INFO] Zeppelin: Hive interpreter ......................... SUCCESS [ 2.094 s]
[INFO] Zeppelin: Tajo interpreter ......................... SUCCESS [ 1.971 s]
[INFO] Zeppelin: JDBC interpreter ......................... SUCCESS [ 1.922 s]
[INFO] Zeppelin: web Application .......................... SUCCESS [ 45.557 s]
[INFO] Zeppelin: Server ................................... SUCCESS [ 4.682 s]
[INFO] Zeppelin: Packaging distribution ................... SUCCESS [ 0.439 s]

But when I execute "bin/zeppelin-daemon.sh start" and try to add the interpreter in http://localhost:8080/ I can not see the option "jdbc" in the dropdown "Interpreter".

[1teed]

@sylvinho81

Thank you for your interest.

I think you seem to be not latest source.
Please update to the latest sources.

Please check this update.

[sylvinho81]

Thanks @1teed. I had not seen the latest changes

[felixcheung]

Since this is using java.sql.DriverManager, it should support all JDBC driver with the same code, and not just mysql, correct?

[1teed]

@felixcheung are correct.
Using java.sql.DriverManager is support all JDBC driver with the same code.
(such as jdbc/src/main/java/org/apache/zeppelin/jdbc/MysqlConnection.java : Line 73)

But every JDBC driver name and url pattern is different.

[jmrr]

@1teed Thanks for the contribution. Tested with a centos-zeppeling docker image and works fine! Looking forward to the merge!

[Leemoonsoo]

version should be 0.5.0-incubating-SNAPSHOT
and line 31, too. (check #59)

[Leemoonsoo]

@hyonaldobe4b52b changes the mysql dependency version. but it is still GPLv2 License.

In my understanding, If code is depends on some GPL (depends means more like can not functioning without it), it can not be included.
And in our case, mysql driver is GPL and this interpreter looks like pretty much depends on mysql driver.

If i'm right, I think it's better to make JdbcInterpreter completely independent to mysql.
And then we can make something similar to dependency loader of sparkinterpreter. so any jdbc driver could be loaded on the fly Instead of adding them in pom.xml.

Any ideas?

[jmrr]

Are temporary tables possible with the interpreter connector? I quickly tried one and got Can not issue data manipulation statements with executeQuery().

Apart from that everything else runs smoothly.

Thanks!

[ZhijieWang]

@1teed I added postgresql support and create a merge request to your repo.

Limitation of this interpreter -- may crush zeppelin when running large query.

[jongyoul]

@Leemoonsoo Concerning licenses, I've heard that GPL shouldn't be included in apache projects, and known that some project in Apache also doesn't have a mysql-connector-java packages due to same reason. I think JdbcInterpreter looks better and general. @1teed How about you?

[riomario1]

Looks good, when are we merging it?

[jmrr]

+1 please go ahead and merge into master

On Wed, Jun 10, 2015 at 6:40 PM, riomario1 notifications@github.com wrote:

Looks good, when are we merging it?

—
Reply to this email directly or view it on GitHub
#60 (comment)
.

[jmrr]

What happened to this? It hasn't been merged to master yet and I'm having problems merging it locally as the master branch is starting to diverge...

Thanks

[rquast]

Is this getting merged still? Just tried it out and found that it could do with some default settings for the interpreter (jdbc - default it to mysql, host, user, password, port). Works great - very useful.

[Leemoonsoo]

I'm not sure about merging it, while this implementation depends on mysql-connector-java that distributed in GPL 2 license. Please correct me if i'm wrong.

[tdunning]

Yes. The GPL license makes mysql-connector be category X. We cannot ship a release with that component, even unmodified.

[martin-g]

Do you pack mysql-connector.jar within the release?
If YES then there is a problem.
If the user has to provide the jar herself (e.g. by executing mvn install locally) then everything is fine.

[riomario1]

Hi,

How do I use MySQL with zeppelin?

Please provide steps

Thank you

[Leemoonsoo]

Do you pack mysql-connector.jar within the release?

I think answer currently is yes, although binary packages are provided as a favor.

Exclude mysql interpreter by default from the build, but let user enabling with some special flag, will this solve the license issue?

[martin-g]

At http://apache.cbox.biz/incubator/zeppelin/0.5.0-incubating/ (my neareast Apache mirror) I see two binary downloads and one source download.
If you don't include mysql-connector.jar in the binaries then everything is fine from legal point of view.
People wanting to use the MySQL interpreter should build Zeppelin from source (i.e. should use the third download: zeppelin-0.5.0-incubating.tgz), possible with some Maven profile that enables/activates MySQL interpreter.
Zeppelin's distributions should not bring mysql-connector.jar with them. Maven's <dependency>...</dependency> and code like import com.mysql.* doesn't violate the rules, so it is safe to have it in Apache sources.

[neptun2000]

When do you planning to release MySQL interpreter ?
Thanks

[riomario1]

Mysql is the most awaited feature :)

Sent from my iPhone

On Aug 29, 2015, at 4:18 PM, neptun2000 notifications@github.com wrote:

When do you planning to release MySQL interpreter ?
Thanks

—
Reply to this email directly or view it on GitHub.

[jongyoul]

@1teed Hi, do you still have an interest to contribute this feature? I've found that your PR is the first one for supporting JDBC thus I think it's important to ask your opinion.

[jongyoul]

@1teed Now, I will try to contact @andrescelis to accept jdbc interpreter. I hope you would understand it. And please check the mailing(https://www.mail-archive.com/dev@zeppelin.incubator.apache.org/msg05504.html) and issue(https://issues.apache.org/jira/browse/ZEPPELIN-439)

[jongyoul]

@1teed please close this PR.