Over a million developers have joined DZone.

Memory Problems: How to Identify a Java Memory Leak

Garbage collection can be garbage sometimes. Learn how to identify Java memory leaks, and drastically improve performance.

· Performance Zone

Download Forrester’s “Vendor Landscape, Application Performance Management” report that examines the evolving role of APM as a key driver of customer satisfaction and business success, brought to you in partnership with BMC.

Every Java developer knows the story: Java harnesses the power of garbage collection to automatically keep application memory clean and healthy. I’ll tell you what though, this is only half the story. While helpful, garbage collection performed automatically in the background isn’t a sure fire way of protecting your app from memory leaks.

The Java Garbage Collector at work:

You know that your app is in trouble when you come across the following message:

java.lang.OutOfMemoryError: Java heap space

If the issue is really too little allocated memory, you can increase the available memory for your application as follows:

java -Xms<initial heap size> -Xmx<maximum heap size>

However, if your problem is a memory leak, then this approach only shifts the problem.

Identifying Memory Leaks

You should check your app for memory leaks even before putting it into production. As an initial indicator, check the garbage collection metrics.

Image title

Java memory leak

If memory usage returns to the same level following GC, all is well. However, if memory rises continuously, you have a problem. The screenshot above shows a provoked memory leak. The code has been taken from this gist on github.com. While in reality, memory consumption would not rise as linear as in this example (Old Gen is the graph of interest), it clearly shows that GC suspension times or Eden and Survivor space usage do not necessarily act as indicators for memory leaks.

Nailing Down the Problem

You have a few options for identifying the cause of such memory issues (for example, JVisualVM and jStat). These tools come with the JDK, so you can jump in anytime and begin investigating. In addition to identifying some frequently used internal Java classes, you may be able to identify some of your own classes.

Optimizing and Tuning

You probably don’t typically care much about how your memory settings are configured—at least as long as garbage collection isn’t impacting performance. This can be a problem because memory issues don’t only manifest themselves in the form of out-of-memory errors and memory leaks. Upcoming memory issues might manifest themselves in the form of high garbage collection times (but not always).

Image title

Garbage collection takes up to 16% of available CPU usage.

Heap Size Set to Small

Heap size set too low leads to increased GC activity because GC must then be executed more often. As you might imagine, increased GC usage leads to increased CPU usage and—as JVMs tend to freeze during GC—often bad performance. GCs tend to be short, but frequent.

Heap Size Set to Large

Heap size set too large manifests in the form of long GC times. GC isn’t performed too often, but once triggered, it stops your VM for a considerable amount of time.

Now consider how a memory leak contributes to such a situation: a memory leak combines the problems of too small heap size with the problems of too large a heap size: GC is triggered frequently and it takes a long time before the JVM finally crashes with an out-of-memory error.

GC Versions

The garbage collector has changed a lot since Java 6. Java 7 introduced the G1GC as an alternative to the stock CMS GC (concurrent mark and sweep) which is slated to become the default for Java 9. Java 8 dropped the PermGen space. The data stored there previously is now stored in either native memory or on the stack.

Though Ruxit takes these differences into account automatically, it’s a good idea for you to understand the subtle differences:

A Final Thought

In addition to displaying JVM stats, Ruxit really shines because it correlates JVM metrics with other key indicators. No other tool makes it so easy to compare GC suspension times with CPU usage, network traffic outages, and more. This makes it easy to think outside the box—unusual JVM behavior may be the result of misbehaving GC, but this isn’t always the case.

See Forrester’s Report, “Vendor Landscape, Application Performance Management” to identify the right vendor to help IT deliver better service at a lower cost, brought to you in partnership with BMC.

performance,memory,RAM,memory leak,memory leaks,java,garbage collection,garbage collector

Published at DZone with permission of Martin Goodwell, DZone MVB. See the original article here.

Opinions expressed by DZone contributors are their own.

The best of DZone straight to your inbox.

Please provide a valid email address.

Thanks for subscribing!

Awesome! Check your inbox to verify your email so you can start receiving the latest in tech news and resources.

{{ parent.title || parent.header.title}}

{{ parent.tldr }}

{{ parent.urlSource.name }}