How to make safe frequent DataSource switches for AbstractRoutingDataSource?
- by serg555
I implemented Dynamic DataSource Routing for Spring+Hibernate according to this article. I have several databases with same structure and I need to select which db will run each specific query.
Everything works fine on localhost, but I am worrying about how this will hold up in real web site environment. They are using some static context holder to determine which datasource to use:
public class CustomerContextHolder {
private static final ThreadLocal<CustomerType> contextHolder =
new ThreadLocal<CustomerType>();
public static void setCustomerType(CustomerType customerType) {
Assert.notNull(customerType, "customerType cannot be null");
contextHolder.set(customerType);
}
public static CustomerType getCustomerType() {
return (CustomerType) contextHolder.get();
}
public static void clearCustomerType() {
contextHolder.remove();
}
}
It is wrapped inside some ThreadLocal container, but what exactly does that mean? What will happen when two web requests call this piece of code in parallel:
CustomerContextHolder.setCustomerType(CustomerType.GOLD);
//<another user will switch customer type here to CustomerType.SILVER in another request>
List<Item> goldItems = catalog.getItems();
Is every web request wrapped into its own thread in Spring MVC? Will CustomerContextHolder.setCustomerType() changes be visible to other web users? My controllers have synchronizeOnSession=true.
How to make sure that nobody else will switch datasource until I run required query for current user?
Thanks.
