Description

quarkus.hibernate-orm.multitenant-schema-datasource seems to have the exact same purpose as quarkus.hibernate-orm.datasource: it sets the (single) datasource to be used for a given persistence unit.

The only difference I could see is that quarkus.hibernate-orm.multitenant-schema-datasource can only be used in a multi-tenancy scenario. All I can find in the documentation is this example, which isn't clear about what the purpose of the property is exactly:

# You could use a non-default datasource by using the following setting
# quarkus.hibernate-orm.multitenant-schema-datasource=other

There are apparently no tests involving the property multitenant-schema-datasource.

Also, judging from the code, unless multitenant-schema-datasource is set, the datasource is duplicated for each tenant:

    private static AgroalDataSource tenantDataSource(String dataSourceName, String tenantId, MultiTenancyStrategy strategy,
            String multiTenancySchemaDataSourceName) {
        if (strategy != MultiTenancyStrategy.SCHEMA) {
            return Arc.container().instance(AgroalDataSource.class, new DataSource.DataSourceLiteral(tenantId)).get();
        }

        if (multiTenancySchemaDataSourceName == null) {
            AgroalDataSource dataSource = getDataSource(dataSourceName);
            // HERE we duplicate the datasource (?)
            return createFrom(dataSource.getConfiguration());
        }

        return getDataSource(multiTenancySchemaDataSourceName);
    }

However, this difference is not documented, seems buggy (the created datasource is never closed) and I'm not sure what the point of copying the datasource even is?

Implementation ideas

We should either clarify the documentation, or fix the behavior and deprecate the property multitenant-schema-datasource for removal.