SmartPool is a connection-pooling component modeled on connection pooling
features provided by an Application Server. SmartPool makes an attempt to
solve critical issues like connection leaks, connection blocking, open JDBC
objects like Statements, PreparedStatements etc. Features of SmartPool include
support for multiple pools, automatic closing of associated JDBC objects,
detect connection leaks based on configurable time-outs, track connection
usage, wrap SmartPool to an existing pool, monitor run time status of the
pools apart from conventional pooling support.
Using the PoolManager:
PoolManager is an interface defining the behavior of the class that
manages each Pool. You can load SmartPool component by using the following:
PoolManager pm = new PoolManagerImpl("/home/ssshetty/pool-config.xml");
Once the PoolManger is instantiated you can call the methods on the
PoolManager interface with the following:
String poolName = "MyPool";
Connection conn = pm.getConnection(poolName);
For more details refer to Java Documentation of PoolManager Interface.
Note: You are creating an Instance of a PoolManager, which is not
Singleton by itself, and thus multiple PoolManager instances would
result in multiple pools.
SmartPoolFactory provides a Singleton access to SmartPool. This is a
Singleton Wrapper to the PoolManagerImpl. This is the recommended way
to load the Pool.
SmartPoolFactory smp = new SmartPoolFactory("/home/pool-config.xml");
This would create a Singleton object smp in the memory. Static methods on this
object can be invoked from any class loaded within the same JVM.
Connection con = SmartPoolFactory.getConnection();
Note: Static methods on SmartPoolFactory can only be used after it is
initialized with a configuration file as shown above. It is recommended to
load this SmartPoolFactory on your application start-up (like the
start-up servlet in a web application).
For more details refer to Java Documentation of SmartPoolFactory.
Accessing Connections from the
Connection can be accessed from the PoolManager/SmartPoolFactory
depending on the method you have used to load the pool. The following examples
are applicable when you use SmartPoolFactory.
You can draw connections from the pool is the following way:
Connection conn = SmartPoolFactory.getConnection();
This fetches a connection from the default pool, with owner set to N/A i.e.
Default Pool: Default pool is one where is set to default-pool="true"
in the configuration file. In a simple application where only one pool is
required and no ownership is to be tracked, developers need not provide the
pool name each time they take a connection from the pool. At most one pool can
be marked as a default pool.
Owner: Owner is the identity of the user/class drawing the connection
from the pool. Ownership is used for debugging while detecting connection
leaks to exactly identify the owners of connections. Preferably this should be
a combination of the class and method that is drawing the connection since
this would directly help in attacking and solving the problem of Connection
Anonymous Connection: A connection drawn from the pool without
specifying the owner name is an anonymous connection. It is not possible to
keep track of ownerships if anonymous connections are allowed.
To disallow anonymous connections set allow-anonymous-connections="false" in
the configuration file.
Other methods to get Connections are:
Connection conn = SmartPoolFactory.getConnection("poolName");
Here poolName is the pool name.
Connection conn = SmartPoolFactory.getConnection("poolName", “TestSmartPool:getMyConnection”);
Here TestSmartPool:getMyConnection is the owner name.
Releasing Connection to the pool:
To release connection to the Pool just call close on the method.
Connection conn = SmartPoolFactory.getConnection("poolName")
This will return the connections to the pool.
Following points are worth noting when a connection is released.
1. If auto-commit is set to false for the connection, SmartPool will call
rollback on that method and set auto-commit to true. This is to destroy the
existing state of the connection before it is made available to others.
2. If auto-close="true" in the configuration file, all the associated
Statements, PreparedStatements, and CallableStatements will be closed when the
connection is released to the pool.
Leak detection is time-out based.
To set a time-interval of 100 seconds for which a class can hold the
connection without being considered as leak set leak-time-out="100".
A Connection Leak is said to have occurred when a class is holding on
to a connection for more than leak-time-out.
A leak detector thread will poll all used connections every 100 seconds.
As soon as the leak detector Poll thread finds out a Connection leak, it will
notify all the registered connection leak listeners.
Connection leak listener is a class implementing the ConnectionLeakListener
interface. Connection leak listener class receives a notification of a leak
through the following call back method:
public void connectionTimeOut(ConnectionLeakEvent cle)
Where ConnectionLeakEvent is the class encapsulating the information
about the leak.
See Java documentation of ConnectionLeakEvent,
ConnectionLeakListener for more info.
You can register a ConnectionLeakLister by calling the following method
You can also provide a default listener in the configuration file. For e.g.:
PoolManager will load this listener on load and it will be able to
listen to leaks right from the instance PoolManager object is created.
NOTE: The connection leak event will also provide you with access to the
actual Connection object which you can be returned back to the pool by calling
close() in it, however be sure of what you are doing, because this may result
in functionality bugs which would be very difficult to debug.
Monitoring Pool Status
Monitoring Interfaces allows you to keep track how the connections are being
used, who is blocking up the connections, detect leaks if any, thus helping in
identifying and attacking problems when the application/system crashes because
of excessive connections or tables get locked etc.
To monitor the Pools you can use the Following method:
PoolMonitor pm = SmartPoolFactory.getPoolMonitor("Sachin");
PoolMonitor is the interface defining methods to monitor the current
status of the pool.
To get the Configuration of the pool, use the following method:
ConfigMonitor cm = pm.getConfigMonitor();
ConfigMonitor provides access to pool configuration.
To get information on the Connections being used:
Vector v = pm.getConnectionsInUse();
This will return a vector of ConnectionMonitor providing all the
information about the connection in use.
For more information see Java Documentation of PoolMonitor,
Automatic closing of associated Statements, PreparedStatements, and
If auto-close="true" in the configuration file then all the
Statements, PreparedStatements, and CallableStatements associated with the
connection will be closed when the connection is released back to the pool.
Wrapping SmartPool to an
Wrapping SmartPool to an existing pool implies that SmartPool will not draw
raw connections to the database; it would rather draw connections from another
connection pool, for e.g. DataSource of an Application Server.
In this case SmartPool does not maintain any pool, it simply delegates the
connection requests to the parent pool. SmartPool however keeps a track of
connections in use, and thus can be used for leak detecting, automatic closing
of associated Statements, PreparedStatements, CallableStatements objects.
SmartPool can be wrapped to an existing pool with the help of the interface
ConnectionProvider. ConnectionProvider provides two methods:
public Connection getConnection() throws Exception
This method should return a connection from the pool to which SmartPool is
public void returnConnection(Connection conn) throws Exception
This method should return the connection conn back to the parent pool.
For more information see Java Documentation of ConnectionProvider.
To wrap smart pool to an existing pool, you need to write an implementation of
ConnectionProvider and provide the fully qualified class name in the
For e.g. If the class implementing ConnectionProvider is
com.your_company.your_project.SampleConnectionProviderImpl, then in the
configuration file, you will have to put the following entry:
Forced recovery of recently used connections
The above specifies the maximum amount of time for which a consumer can
hold on to a connection, without using it. hence if the poll thread finds any
connections that have not been used for more than 100 seconds, it will
take the connection back, and any attempt to use the connection
afterwards will result in StaleConnectionException.
DTD for configuration file
Sample configuration file
Consumer: Consumer is a class/component that draws connections from the
Default Pool: Default pool is the pool where you have said
default-pool="true" in the configuration file. Thus in a simple application
where only one pool is required and no ownership is to be tracked, developers
need not provide the pool name each time they take a connection from the pool.
Only one pool can be marked as a default pool at the most.
Owner: Owner is an identity of the user/class drawing the connection
from the pool. This is used for debugging while detecting connection leaks so
as to exactly identify who is blocking the connections. Preferably this should
be a combination of the class and method that is drawing the connection as
this would directly help in attacking and solving the problem of Connection
Anonymous Connection: An anonymous connection is a connection drawn
from the pool without specifying the owner name. Hence not possible to keep
track of where the connection is being blocked.
When allow-anonymous-connections="false" in the configuration file, anonymous
connections are not allowed.
Connection Leak: When a Consumer holds on to a connection for more than
the time specified in leak-time-out in configuration file, a connection leak
is said to have occurred and the Consumer is the owner of the connection and
is responsible for the connection leak.