3 Tips to Help Avoid the Hibernate Flush

When using Grails, the default behavior of Hibernate is to do a flush before queries, at the end of requests, and pretty much whenever Hibernate feels like it. So here are three tips that can help you avoid some of the flushing and/or at the very least decrease the number of database connections that Hibernate attempts to get:
  1. Use read() function when possible: If you are only going to read some field of a domain object and will make no changes to it, then just use object.read(), instead of the get() method
  2. Make sure your services are only transactional if they need to be, so if you are not updating, saving, or deleting records in the db chances are you do not need your Service to be transactional
  3. If you do have some transactional logic, put it inside a domain.withTransaction {} closure instead of making the entire function or class transactional

Configuration of DataSource.groovy for Highly Concurrent Grails Application

In developing a highly concurrent Grails Application, I ran into an issue where many connections were being created to the database and were subsequently not released. After reading several blog posts it seems that the library used to handle the connection pooling was not designed for highly concurrent systems. I found that using the following setup in your DataSource.groovy file eliminates the stale connection issue and allows a high level of concurrency.

dataSource {
pooled = true
driverClassName = "com.mysql.jdbc.Driver"
dialect = "org.hibernate.dialect.MySQL5InnoDBDialect"
properties {
maxActive = 50
maxIdle = 10
initialSize = 10
minEvictableIdleTimeMillis = 10000
timeBetweenEvictionRunsMillis = 20000
maxWait = 10000
validationQuery = "/* ping */"
testWhileIdle = true
numTestsPerEvictionRun = 3
testOnBorrow = true

Hope this can help someone else.