Generic JDBC Interpreter for Apache Zeppelin

Overview

JDBC interpreter lets you create a JDBC connection to any data sources seamlessly.

Inserts, Updates, and Upserts are applied immediately after running each statement.

By now, it has been tested with:

If you are using other databases not in the above list, please feel free to share your use case. It would be helpful to improve the functionality of JDBC interpreter.

Create a new JDBC Interpreter

First, click + Create button at the top-right corner in the interpreter setting page.

Fill Interpreter name field with whatever you want to use as the alias(e.g. mysql, mysql2, hive, redshift, and etc..). Please note that this alias will be used as %interpreter_name to call the interpreter in the paragraph. Then select jdbc as an Interpreter group.

The default driver of JDBC interpreter is set as PostgreSQL. It means Zeppelin includes PostgreSQL driver jar in itself. So you don't need to add any dependencies(e.g. the artifact name or path for PostgreSQL driver jar) for PostgreSQL connection. The JDBC interpreter properties are defined by default like below.

Name Default Value Description
common.max_count 1000 The maximun number of SQL result to display
default.driver org.postgresql.Driver JDBC Driver Name
default.password The JDBC user password
default.url jdbc:postgresql://localhost:5432/ The URL for JDBC
default.user gpadmin The JDBC user name
default.precode Some SQL which executes every time after initialization of the interpreter (see Binding mode)
default.statementPrecode SQL code which executed before the SQL from paragraph, in the same database session (database connection)
default.completer.schemaFilters Сomma separated schema (schema = catalog = database) filters to get metadata for completions. Supports '%' symbol is equivalent to any set of characters. (ex. prod_v_%,public%,info)
default.completer.ttlInSeconds 120 Time to live sql completer in seconds (-1 to update everytime, 0 to disable update)

If you want to connect other databases such as Mysql, Redshift and Hive, you need to edit the property values. You can also use Credential for JDBC authentication. If default.user and default.password properties are deleted(using X button) for database connection in the interpreter setting page, the JDBC interpreter will get the account information from Credential.

The below example is for Mysql connection.

The last step is Dependency Setting. Since Zeppelin only includes PostgreSQL driver jar by default, you need to add each driver's maven coordinates or JDBC driver's jar file path for the other databases.

That's it. You can find more JDBC connection setting examples(Mysql, MariaDB, Redshift, Apache Hive, Presto/Trino, Impala, Apache Kyuubi, Apache Phoenix, and Apache Tajo) in this section.

JDBC Interpreter Datasource Pool Configuration

The Jdbc interpreter uses the connection pool technology, and supports users to do some personal configuration of the connection pool. For example, we can configure default.validationQuery='select 1' and default.testOnBorrow=true in the Interpreter configuration to avoid the "Invalid SessionHandle" runtime error caused by Session timeout when connecting to HiveServer2 through JDBC interpreter.

The Jdbc Interpreter supports the following database connection pool configurations:

Property Name Default Description
testOnBorrow false The indication of whether objects will be validated before being borrowed from the pool. If the object fails to validate, it will be dropped from the pool, and we will attempt to borrow another.
testOnCreate false The indication of whether objects will be validated after creation. If the object fails to validate, the borrow attempt that triggered the object creation will fail.
testOnReturn false The indication of whether objects will be validated before being returned to the pool.
testWhileIdle false The indication of whether objects will be validated by the idle object evictor (if any). If an object fails to validate, it will be dropped from the pool.
timeBetweenEvictionRunsMillis -1L The number of milliseconds to sleep between runs of the idle object evictor thread. When non-positive, no idle object evictor thread will be run.
maxWaitMillis -1L The maximum number of milliseconds that the pool will wait (when there are no available connections) for a connection to be returned before throwing an exception, or -1 to wait indefinitely.
maxIdle 8 The maximum number of connections that can remain idle in the pool, without extra ones being released, or negative for no limit.
minIdle 0 The minimum number of connections that can remain idle in the pool, without extra ones being created, or zero to create none.
maxTotal -1 The maximum number of active connections that can be allocated from this pool at the same time, or negative for no limit.
validationQuery show database The SQL query that will be used to validate connections from this pool before returning them to the caller. If specified, this query MUST be an SQL SELECT statement that returns at least one row. If not specified, connections will be validation by calling the isValid() method.

More properties

There are more JDBC interpreter properties you can specify like below.

Property Name Description
common.max_result Max number of SQL result to display to prevent the browser overload. This is common properties for all connections
zeppelin.jdbc.auth.type Types of authentications' methods supported are SIMPLE, and KERBEROS
zeppelin.jdbc.principal The principal name to load from the keytab
zeppelin.jdbc.keytab.location The path to the keytab file
zeppelin.jdbc.auth.kerberos.proxy.enable When auth type is Kerberos, enable/disable Kerberos proxy with the login user to get the connection. Default value is true.
default.jceks.file jceks store path (e.g: jceks://file/tmp/zeppelin.jceks)
default.jceks.credentialKey jceks credential key
zeppelin.jdbc.interpolation Enables ZeppelinContext variable interpolation into paragraph text. Default value is false.
zeppelin.jdbc.maxConnLifetime Maximum of connection lifetime in milliseconds. A value of zero or less means the connection has an infinite lifetime.

You can also add more properties by using this method. For example, if a connection needs a schema parameter, it would have to add the property as follows:

name value
default.schema schema_name

How to use

Run the paragraph with JDBC interpreter

To test whether your databases and Zeppelin are successfully connected or not, type %jdbc_interpreter_name(e.g. %mysql) at the top of the paragraph and run show databases.

%jdbc_interpreter_name

show databases

If the paragraph is FINISHED without any errors, a new paragraph will be automatically added after the previous one with %jdbc_interpreter_name. So you don't need to type this prefix in every paragraphs' header.

Multiple SQL statements

You can write multiple sql statements in one paragraph, just separate them with semi-colon. e.g

%jdbc_interpreter_name

USE zeppelin_demo;

CREATE TABLE pet (name VARCHAR(20), owner VARCHAR(20),
       species VARCHAR(20), sex CHAR(1), birth DATE, death DATE);

SQL Comment

2 kinds of SQL comments are supported:

  • Single line comment start with --
  • Multiple line comment around with /* ... */
%jdbc_interpreter_name

-- single line comment
show tables;
/* multiple 
   line 
   comment
 */
select * from test_1;

Apply Zeppelin Dynamic Forms

You can leverage Zeppelin Dynamic Form inside your queries. You can use both the text input and select form parametrization features.

Run SQL Continuously

By default, sql statements in one paragraph are executed only once. But you can run it continuously by specifying local property refreshInterval (unit: milli-seconds), So that the sql statements are executed every interval of refreshInterval milli-seconds. This is useful when your data in database is updated continuously by external system, and you can build dynamic dashboard in Zeppelin via this approach.

e.g. Here we query the mysql which is updated continuously by other external system.

Usage precode

You can set precode for each data source. Code runs once while opening the connection.

Properties

An example settings of interpreter for the two data sources, each of which has its precode parameter.

Property Name Value
default.driver org.postgresql.Driver
default.password 1
default.url jdbc:postgresql://localhost:5432/
default.user postgres
default.precode set search_path='test_path'
default.driver com.mysql.jdbc.Driver
default.password 1
default.url jdbc:mysql://localhost:3306/
default.user root
default.precode set @v=12
Usage

Test of execution precode for each data source.

%jdbc

show search_path

Returns value of search_path which is set in the default jdbc (use postgresql) interpreter's default.precode.

%mysql

select @v

Returns value of v which is set in the mysql interpreter's default.precode.

Examples

Here are some examples you can refer to. Including the below connectors, you can connect every databases as long as it can be configured with it's JDBC driver.

Postgres

Properties
Name Value
default.driver org.postgresql.Driver
default.url jdbc:postgresql://localhost:5432/
default.user mysql_user
default.password mysql_password

Postgres JDBC Driver Docs

Dependencies
Artifact Excludes
org.postgresql:postgresql:42.3.3

Maven Repository: org.postgresql:postgresql

Mysql

Properties
Name Value
default.driver com.mysql.jdbc.Driver
default.url jdbc:mysql://localhost:3306/
default.user mysql_user
default.password mysql_password

Mysql JDBC Driver Docs

Dependencies
Artifact Excludes
mysql:mysql-connector-java:5.1.38

Maven Repository: mysql:mysql-connector-java

MariaDB

Properties
Name Value
default.driver org.mariadb.jdbc.Driver
default.url jdbc:mariadb://localhost:3306
default.user mariadb_user
default.password mariadb_password

MariaDB JDBC Driver Docs

Dependencies
Artifact Excludes
org.mariadb.jdbc:mariadb-java-client:1.5.4

Maven Repository: org.mariadb.jdbc:mariadb-java-client

Redshift

Properties
Name Value
default.driver com.amazon.redshift.jdbc42.Driver
default.url jdbc:redshift://your-redshift-instance-address.redshift.amazonaws.com:5439/your-database
default.user redshift_user
default.password redshift_password

AWS Redshift JDBC Driver Docs

Dependencies
Artifact Excludes
com.amazon.redshift:redshift-jdbc42:2.1.0.18

Maven Repository: com.amazon.redshift:redshift-jdbc42

Apache Hive

Zeppelin just connect to hiveserver2 to run hive sql via hive jdbc. There are 2 cases of connecting with Hive:

  • Connect to Hive without KERBEROS
  • Connect to Hive with KERBEROS

Each case requires different settings.

Connect to Hive without KERBEROS

In this scenario, you need to make the following settings at least. By default, hive job run as user of default.user. Refer impersonation if you want hive job run as the Zeppelin login user when authentication is enabled.

Name Value
default.driver org.apache.hive.jdbc.HiveDriver
default.url jdbc:hive2://localhost:10000
default.user hive_user
Artifact Excludes
org.apache.hive:hive-jdbc:2.3.4
Connect to Hive with KERBEROS

In this scenario, you need to make the following settings at least. By default, hive job run as user of client principal (zeppelin.jdbc.principal). Refer impersonation if you want hive job run as the Zeppelin login user when authentication is enabled.

Name Value
default.driver org.apache.hive.jdbc.HiveDriver
default.url jdbc:hive2://emr-header-1:10000/default;principal={hive_server2_principal}
zeppelin.jdbc.auth.type KERBEROS
zeppelin.jdbc.keytab.location keytab of client
zeppelin.jdbc.principal principal of client
Artifact Excludes
org.apache.hive:hive-jdbc:2.3.4
org.apache.hive:hive-exec:2.3.4

Maven Repository : org.apache.hive:hive-jdbc

Impersonation

When Zeppelin server is running with authentication enabled, then the interpreter can utilize Hive's user proxy feature i.e. send extra parameter for creating and running a session ("hive.server2.proxy.user=": "${loggedInUser}"). This is particularly useful when multiple users are sharing a notebook.

To enable this set following:

  • default.proxy.user.property as hive.server2.proxy.user

See User Impersonation in interpreter for more information.

Sample configuration
Name Value
default.driver org.apache.hive.jdbc.HiveDriver
default.url jdbc:hive2://hive-server-host:2181/;serviceDiscoveryMode=zooKeeper;zooKeeperNamespace=hiveserver2
default.proxy.user.property default.server2.proxy.user
zeppelin.jdbc.auth.type SIMPLE

See Hive Interpreter for more properties about Hive interpreter.

Presto/Trino

Properties

Name Value
default.driver io.prestosql.jdbc.PrestoDriver
default.url jdbc:presto://presto-server:9090/hive
default.user presto_user

Trino JDBC Driver Docs
Presto JDBC Driver Docs

Dependencies

Artifact Excludes
io.prestosql:presto-jdbc:350

Impala

Properties

Name Value
default.driver org.apache.hive.jdbc.HiveDriver
default.url jdbc:hive2://emr-header-1.cluster-47080:21050/;auth=noSasl

Dependencies

Artifact Excludes
org.apache.hive:hive-jdbc:2.3.4

Impala JDBC Driver Docs

Dependencies

Artifact Excludes
io.prestosql:presto-jdbc:350

Apache Kyuubi

Zeppelin connect to Kyuubi to run sql via KyuubiHiveDriver. There are 2 cases of connecting with Kyuubi:

  • Connect to Kyuubi without KERBEROS
  • Connect to Kyuubi with KERBEROS

Each case requires different settings.

Connect to Kyuubi without KERBEROS

In this scenario, you need to make the following settings at least. Kyuubi engine run as user of default.user.

Properties

Name Value
default.driver org.apache.kyuubi.jdbc.KyuubiHiveDriver
default.url jdbc:hive2://kyuubi-server:10009

Dependencies

Artifact Excludes
org.apache.kyuubi:kyuubi-hive-jdbc-shaded:1.6.1-incubating
org.apache.hive:hive-jdbc:3.1.2
Connect to Kyuubi with KERBEROS

In this scenario, you need to make the following settings at least. Kyuubi engine run as user of client principal (zeppelin.jdbc.principal).

Properties

Name Value
default.driver org.apache.kyuubi.jdbc.KyuubiHiveDriver
default.url jdbc:hive2://kyuubi-server:10009/default;principal={kyuubi_principal}
zeppelin.jdbc.auth.type KERBEROS
zeppelin.jdbc.keytab.location keytab of client
zeppelin.jdbc.principal principal of client

Dependencies

Artifact Excludes
org.apache.kyuubi:kyuubi-hive-jdbc-shaded:1.6.1-incubating
org.apache.hive:hive-jdbc:3.1.2

Apache Phoenix

Phoenix supports thick and thin connection types:

Use the appropriate default.driver, default.url, and the dependency artifact for your connection type.

Thick client connection

Properties
Name Value
default.driver org.apache.phoenix.jdbc.PhoenixDriver
default.url jdbc:phoenix:localhost:2181:/hbase-unsecure
default.user phoenix_user
default.password phoenix_password
Dependencies
Artifact Excludes
org.apache.phoenix:phoenix-core:4.4.0-HBase-1.0

Maven Repository: org.apache.phoenix:phoenix-core

Thin client connection

Properties
Name Value
default.driver org.apache.phoenix.queryserver.client.Driver
default.url jdbc:phoenix:thin:url=http://localhost:8765;serialization=PROTOBUF
default.user phoenix_user
default.password phoenix_password
Dependencies

Before Adding one of the below dependencies, check the Phoenix version first.

Artifact Excludes Description
org.apache.phoenix:phoenix-server-client:4.7.0-HBase-1.1 For Phoenix 4.7
org.apache.phoenix:phoenix-queryserver-client:4.8.0-HBase-1.2 For Phoenix 4.8+

Maven Repository: org.apache.phoenix:phoenix-queryserver-client

Apache Tajo

Properties
Name Value
default.driver org.apache.tajo.jdbc.TajoDriver
default.url jdbc:tajo://localhost:26002/default

Apache Tajo JDBC Driver Docs

Dependencies
Artifact Excludes
org.apache.tajo:tajo-jdbc:0.11.2

Maven Repository: org.apache.tajo:tajo-jdbc

Object Interpolation

The JDBC interpreter also supports interpolation of ZeppelinContext objects into the paragraph text. The following example shows one use of this facility:

In Scala cell:

z.put("country_code", "KR")
    // ...

In later JDBC cell:

%jdbc_interpreter_name

select * from patents_list where 
priority_country = '{country_code}' and filing_date like '2015-%'

Object interpolation is disabled by default, and can be enabled for all instances of the JDBC interpreter by setting the value of the property zeppelin.jdbc.interpolation to true (see More Properties above). More details of this feature can be found in the Spark interpreter documentation under Zeppelin-Context

Bug reporting

If you find a bug using JDBC interpreter, please create a JIRA ticket.