Spring Batch initialize-schema always will throw error relation "batch_job

I have a Spring Batch application which will be triggered and started by a Kubernetes CronJob once a day and do some operations.

So it is not needed, that the application will run the whole time and idle until the scheduler starts the job.

To initialize the spring batch schema

spring:
  batch:
    jdbc:
      initialize-schema: always

is used.

By doing so, every time the application is triggered to run, the schema will be initialzed again. This is why I get an error in the database logs which says

[380] ERROR:  relation "batch_job_instance" already exists

because the schema will be initialzed always.

I searched for another option than just always or never, but seems nothing else.

I'm using Spring Batch Version 4.3.6 and PostgreSQL.

CodePudding user response：

The spring.batch.jdbc.initialize-schema=always property tells Spring Batch Auto configuration to create the schema everytime the application restarts. This property also silently sets continueOnError= true in the auto-configuration, so even if the error comes it will not fail the application.

You will not observe this behavior locally when with embedded database, if you have any other local DB running on either docker or on system it will throw the error.

It is advisable that in containerize deployment you should set the property to never and do schema creation for batch metadata tables manually.

You can get these DDL script from Spring Batch org/springframework/batch/core/migration package. You can add this to your migration script if you are using Flyway or Liuquibase, or in docker initialization script for your DB service in docker-compose.

Note: Reference Spring Doc here