How do you configure a Camunda process specific datasource?

Question

I have Camunda 7.10 running on Tomcat 9.0.12.

The process engine datasource is configured in

/[tomcat-home]/conf/server.xml

and, by necessity, is available to all deployed web-apps.

I wish to configure a process specific datasource that is only available to that process.

For a normal web app, there is typically a

[tomcat-home]/webapps/[web-app-name]/META-INF/context.xml

configuration file that defines the datasource and the locator code that returns a javax.sql.DataSource using a javax.naming.InitialContext etc

Looking at the deployment of the specific process, there are two META-INF folders

[tomcat-home]\webapps\[process-name]\META-INF
[tomcat-home]\webapps\[process-name]\WEB-INF\classes\META-INF

The first has maven specific items. The second is where the process.xml file resides.

Using proven datasource locator code and putting a proven context.xml file here fails with

javax.naming.NameNotFoundException: Name [jdbc/recruitDS] is not bound in this Context. Unable to find [jdbc].

Searching for this only leads to information on the main Camunda process engine datasource configuration which is well covered in the manual, forums etc.

Kevin Swann Kevin Swann · Accepted Answer · 2019-01-19T12:39:46

The following configuration works for Camunda 7.10 running on Tomcat 9.0.12 accessing SQL Server 2012 using a Maven build project as per the Camunda Java Process Get-Started example.

In the exploded WAR file, in the folder:

[tomcat-home]\webapps\[process-name]\META-INF

there must be a context.xml file with these fields:

<?xml version="1.0" encoding="UTF-8"?>
<Context>
  <Resource
    name="jdbc/myDatasource"
    global="jdbc/myDatasource"
    auth="Container"
    type="javax.sql.DataSource"
    driverClassName="com.microsoft.sqlserver.jdbc.SQLServerDriver"
    url="jdbc:sqlserver://127.0.0.1:1433;databaseName=myDatabaseName"
    username="myUsername"
    password="myPassword"
    maxTotal="50"
    maxIdle="0"
    maxWaitMillis="-1"
    removeAbandonedOnBorrow="true"
    removeAbandonedOnMaintenance="true" />
</Context>

The important field is the global tag - missing this causes the above exception. For a normal JSP/servlet web-app, this field is not necessary as the scope is limited to that web-app. For Camunda, the task listeners etc run in the scope of the main Camunda web-app.

Using a Maven build, it is just a case of adding the META-INF folder & file to the project file structure.

Then, in a Camunda JavaDelegate or TaskListener, it is possible to use the DataSource to obtain a connection in the usual manner via an InitialContext with a URL of the form:

java:/comp/env/jdbc/myDatasource

Using a process specific data-source avoids polluting the main tomcat server.xml configuration file with process specific items.

How do you configure a Camunda process specific datasource?

1 Answers