Using custom Postgres driver

Last modified: 24 March 2025

In certain circumstances, a custom PostgreSQL driver might be required to connect to the Postgres database designated for Datalore internal usage (for example, when RDS is used and requires IAM authentication).

tip

This article is about the Datalore's own internal PostgreSQL database connectivity, and does not affect any of its data source connection. To use a custom JDBC driver for a data source connection, refer to the Add custom JDBC drivers article.

The corresponding JAR file(s) should be placed in the /opt/datalore/tomcat/lib/ext directory. You may either:

  • embed these files into the Datalore image (building the custom images based on jetbrains/datalore-server:2025.2

  • override Datalore startup script located at /opt/datalore/on_start.sh, so these will be downloaded on each server pod startup.

note

Consider increasing the initialDelay in livenessProbe of the server deployment spec, if approach with downloading drivers on startup is chosen.

Once drivers are uploaded, the database connection string should be altered to reference a new driver:

DB_USER='datalore'
DB_PASSWORD=''
DB_URL='jdbc:aws-wrapper:postgresql://jdbc-wrapper-test.anbsra1by3jv.eu-west-1.rds.amazonaws.com/datalore?wrapperPlugins=failover,efm2,iam'

As a final step, the server pod should be restarted as described in this article, so the changes are applied.

Example: Overriding the startup script

  1. Create a Kubernetes manifest defining a new file: 

    apiVersion: v1
kind: ConfigMap
metadata:
  name: datalore-server-on-start
data:
  on_start.sh: >
    #!/bin/bash
    set -e
    (
      cd /opt/datalore/tomcat/lib/ext
      wget https://github.com/aws/aws-advanced-jdbc-wrapper/releases/download/2.5.4/aws-advanced-jdbc-wrapper-2.5.4.jar
      wget https://repo1.maven.org/maven2/software/amazon/awssdk/bundle/2.30.31/bundle-2.30.31.jar
      wget https://jdbc.postgresql.org/download/postgresql-42.7.5.jar
    )

  2. Apply the manifest to the Kubernetes cluster: 

    kubectl apply -f datalore-server-on-start.yaml

  3. Adjust the Datalore server deployment to mount the ConfigMap as a volume: 

    spec:
  containers:
  ...
    volumeMounts:
    - name: datalore-server-on-start
      mountPath: /opt/datalore/on_start.sh
      subPath: on_start.sh
  volumes:
  - name: datalore-server-on-start
    configMap:
      name: datalore-server-on-start

  4. Restart the Datalore server pod, as described in lifecycle management article.