Uncovering Memory Leaks Using NetBeans Profiler

This document is the first of two articles on using the NetBeans Profiler to locate memory leaks in a Java application.

What's a Memory Leak?

"Memory leak is a particular kind of unintentional memory consumption by a computer program where the program fails to release memory when no longer needed" as Wikipedia says. This can happen for any program written in any programming language and Java is no exception.

Many users think that the Java VM releases unused objects from memory automatically, but that's not always the case. There are several scenarios when objects cannot be released and memory leaks occur. One common scenario is when a single object with large memory consumption may be unintentionaly held in memory, for example a Project instance in NetBeans IDE when the project has already been closed. Another common scenario is when small objects are constantly being added to a Collection over time but they are never removed from that Collection, for example caching coordinates in a graphics editor or a game.

How Can I Detect a Memory Leak?

In the first scenario, when a single (or few) object remains in memory after some action, memory leak detection is typically done by comparing memory snapshots taken before the action to snapshots taken after the action and then searching for unnecessary references to that object(s) in a heap walker tool. This technique will be covered in a subsequent article on uncovering memory leaks.

This article will cover the second scenario, when there are many small objects that are continuously accumulating without being released. This type of leak can be more difficult to detect because typically it's not related to any concrete action and doesn't use a noticeable amount of memory until the program has run for a long time. This scenario is dangerous especially with Java EE applications that are expected to run for long periods of time. After some time the leak starts to consume memory and may slow down performance of the application or cause an OutOfMemoryError crash without any visible cause.

Even if your application seems to be working without any problems, before releasing your application it's always a good idea to verify that your application does not contain any potential memory leaks. For this purpose the NetBeans Profiler offers you special metrics which can reliably uncover large amounts of continuously leaking small objects - the Surviving Generations metrics. You can start the application in Monitoring mode without any profiling overhead and watch the graph to see if the application leaks over time.

What Do The Surviving Generations Metrics Mean?

Surviving Generations metrics cannot be easily defined with just one line, so let's make a three-line definition:

• a Generation is a set of instances created within the same GC interval (between two garbage collections)
• a Surviving Generation is a Generation that survives at least one garbage collection. The number of survived garbage collections - the generation's age - is its unique identifier
• Surviving Generations (metrics) value is the number of differerent Surviving Generations that are currently alive on the heap (number of Generations with different generation ages)

Typically there are several long-lived objects (like an application's main JFrame etc.) in an application. The generation's age increases during the application's run time but still represents one or a few Surviving Generations.

Most applications also have short-lived objects which are created very frequently (such as Dimension etc.) but these objects are released very soon, typically within only a few garbage collections. This means that they represent only a few Surviving Generations (with generation's age of 1, 2, 3 etc.).

If we merge the two above cases, the total number of Surviving Generations in typical applications is quite stable. You may have long-lived objects representing, for example, 3 Surviving Generations, and short-lived objects representing, for example, 5 Surviving Generations, which means there will be about 8 Surviving Generations during the application's runtime.

Of course, in some application, for example a Swing application, dialogs and other components are created during run time and as a result the number of Surviving Generations grows a bit. But after some period of time the number of Surviving Generations should become stable because all long-lived objects have already been created and newly-created short-lived objects are periodically being released from the heap.

However, if there is a memory leak in an application which prevents newly-created objects from being released from the heap (for example objects stored in a Collection and never removed), the number of Surviving Generations grows. Why? Because between every two garbage collections a new generation (of leaking objects) is created and it survives each successive garbage collection. During run time the number of Surviving Generations whose generation's age differ by one unit increases, and that's exactly what the Surviving Generations metrics is able to detect regardless of how much memory is wasted by such a leak. That's why using the Surviving Generations metrics can help you discover a memory leak much sooner, before it consumes the entire heap and causes an OutOfMemoryError.

How To Do It With NetBeans Profiler?

1. First download and install the latest stable version of NetBeans Profiler (currently Profiler 5.5) from here. Be sure to read OS and JDK requirements in the release notes document.
2. Then start profiling a project that you want to check for possible memory leaks or attach the NetBeans Profiler to an external application. Refer to the NetBeans Profiler online documentation or check the NetBeans Profiler tutorial to start profiling in Monitoring mode.
   

   Image 1: After invoking Profile Main Project action or Attach Profiler action, the Select Profiling Task dialog appears. Choose the Monitor Application mode. (You don't need to monitor application's threads.)

3. Open the small telemetry graphs (menu Profile/View/Telemetry Overview) or large Memory (GC) graph using the VM Telemetry button in the NetBeans Profiler Control Panel and watch the behavior of the red Surviving Generations metrics line during the application's run time. The longer the application runs in monitoring mode with a realistic workload, the more precise the memory leak analysis will be.
   

   Image 2: VM Telemetry Overview showing how the Surviving Generations value increases after application startup (the exact number of surviving generations is not important). If the value continues to increase during run time, the application is very likely leaking.

   Typically, the Surviving Generations value starts to grow during an application's startup, but after some time it should reach some stable limit and level off. When this happens it indicates that no objects are being created continuously and leaking. If the Surviving Generations line continues to grow all the time (regardless of how quickly), then the application is most likely leaking.

4. If you discover a memory leak using the Surviving Generations metrics during monitoring, the NetBeans Profiler enables you to easily find the code in your application where the leaking objects are created and fix the problem. To find the suspect code, do the following:
   1. Switch the profiling mode to Memory - Liveness with collecting stack traces.
   2. Open the live results and sort the results by Surviving Generations value.
   3. Right-click the class with the continuously growing Surviving Generations and choose Show Allocations Stack Traces in the popup menu.

   Follow these steps to locate all the methods of your code where instances of the class are being created. The method(s) with the highest Surviving Generations value is most likely the cause of the leak. For more detailed information on memory profiling, see the third part of the Profiler tutorial.
   
   

   Image 3: The Live results tab shows that Hashtable entries are continuously being created and not released. This is a potential memory leak even though they don't consume much memory.