How to setup multiple connection pools when multiple datasources are used in Spring Boot?

Here's the approach I had to take in order to get separate pools for each datasource. The following is an implementation of the points that @user3007501 made above.

1. Don't use the DataSourceBuilder, and instead create a org.apache.tomcat.jdbc.pool.DataSource. This will create both the pool, and configure the connections.

   If you need Hikari or Dbcp2 replace the contents of the method createPooledDataSource() below, with the Hikari or Dbcp2 config sections from the original Spring source DataSourceConfiguration.java. The displayed contents of createPooledDataSource() below were stolen from the Tomcat.dataSource() method in the linked file.

2. Add a tomcat configuration section under each of your your datasource configurations in your application.yml
3. Ensure your each of your configuration beans is using the config-name-here.datasource.tomcat (note the .tomcat) property specified in the application.yml and not the config-name-here.datasource without .tomcat
4. Add beans to provide configurations of DataSourceProperties fore each of the datasources
5. use @Qualifier("name of bean from previous step") on your tomcat polling datasource

application.yml

# Primary Datasource
spring:
  datasource:
    username: your-username-for-ds-1
    password: your-password-for-ds-1
    driver-class-name: net.sourceforge.jtds.jdbc.Driver
    tomcat:
      validation-query: select 1
      test-on-borrow: true


myotherdatasource:
  datasource:
    username: your-username-for-ds-2
    password: your-password-for-ds-2
    driver-class-name: net.sourceforge.jtds.jdbc.Driver
    # HERE: make sure you have a tomcat config for your second datasource like below 
    tomcat:
      validation-query: select 1
      test-on-borrow: true

MyCustomDatasourceConfig.java <- your custom config file

The createPooledDataSource() was taken from DataSourceConfiguration.java in the Spring project source.

import org.apache.tomcat.jdbc.pool.DataSource;
import org.springframework.boot.autoconfigure.jdbc.DataSourceProperties;
import org.springframework.boot.context.properties.ConfigurationProperties;
import org.springframework.boot.jdbc.DatabaseDriver;
import org.springframework.context.annotation.Bean;
import org.springframework.context.annotation.Configuration;
import org.springframework.context.annotation.Primary;
import org.springframework.beans.factory.annotation.Qualifier;


@Configuration
public class MyCustomDatasourceConfig {
    @Bean(name = "My-First-Data")
    @Primary
    @ConfigurationProperties(prefix = "spring.datasource.tomcat") 
    // *** NOTE the inclusion of the .tomcat above
    public DataSource primaryDataSource(DataSourceProperties properties) {
        return createPooledDataSource(properties);
    }


    @Bean()
    @Primary
    @ConfigurationProperties(prefix = "spring.datasource")
    public DataSourceProperties dataSourcePropsPrimary() {
        return new DataSourceProperties();
    }


    @Bean(name = "My-Second-Data-Source")
    @ConfigurationProperties(prefix = "myotherdatasource.datasource.tomcat") 
    // *** NOTE the inclusion of the .tomcat above
    public DataSource datasourceOtherConfig(@Qualifier("secondary_ds_prop") DataSourceProperties properties) {
        return createPooledDataSource(properties);
    }

    @Bean(name  = "secondary_ds_prop")
    @ConfigurationProperties(prefix = "myotherdatasource.datasource")
    public DataSourceProperties dataSourcePropsSecondary() {
        return new DataSourceProperties();
    }


    private DataSource createPooledDataSource(DataSourceProperties properties) {
        // Using fully qualified path to the tomcat datasource just to be explicit for the sake of this example
        DataSource dataSource = (org.apache.tomcat.jdbc.pool.DataSource)
                   properties.initializeDataSourceBuilder()
                   .type(org.apache.tomcat.jdbc.pool.DataSource.class).build();
        DatabaseDriver databaseDriver = DatabaseDriver.fromJdbcUrl(properties.determineUrl());
        String validationQuery = databaseDriver.getValidationQuery();
        if (validationQuery != null) {
            dataSource.setTestOnBorrow(true);
            dataSource.setValidationQuery(validationQuery);
        }
        return dataSource;
    }
}

1. Generally, DataSource interface is implemented by pooling libraries to be compatible with frameworks and JavaEE code as universal JDBC connection source which actually works with DB driver.
   SpringBoot autoconfiguration has DataSource initializers for popular pooling libraries. Full list you can find in Spring sources.
   That means all you need to leverage pooling in your project is to add pooling library like Hikari as dependency and configure spring.datasource.* params. Spring will create and configure single pooling DataSource which can be autowired in your code.
2. Another story if you need to create more than one DataSource. SpringBoot autoconfiguration heavily use @ConditionalOnMissingBean annotation to determine cases when default behaviour can be applied. Spring can't create two default datasources because it's ambiguous which one should be used.
   You can find this in spring-boot-autoconfugire module: Spring starts DataSource initialisation logic only if there's no bean of this type in context.
   To use several pools you have to define your own Bean for each pooled DB connection. Spring will notice your DataSource and won't create pool internally. Here's an example.
3. You can find more details about DataSource autoconfiguration here