Understanding Node.js Memory Leaks

September 22 2015

Find out how Node.js memory leaks can affect the performance of your Node apps and how to detect and solve memory leaks in Node.js.

Memory leaks can be a serious problem in Node.js, potentially affecting the performance of your Node apps. Although it might look like a predicament in the back-end is causing the application to fail, the real source of a bug could be a Node.js memory leak. It’s important to understand what memory leaks are, why they occur, and how to detect and solve memory leaks in Node.js, to get ultimately to the bottom of fixing memory leaks.

What Is a Memory Leak?

A memory leak occurs when an application takes over part of a computer’s memory and fails to release it at the appropriate time. Over time, this can lead to the computer running out of memory, causing a decrease in performance. When a Node.js memory leak occurs, the affected application may run slowly or freeze completely. A memory leak can cause a serious performance problem that needs to be diagnosed and solved.

How Does Node.js Manage Memory?

Complications with the V8 garbage collector (Google’s Javascript engine) potentially lead to Node.js memory leaks. V8 is an open source engine that runs Node.js applications, although it was not designed specifically for Node. V8 deals with memory management on behalf of the Node.js applications, making life easier for Node developers because it takes away the manual task of writing code that deals with the management of memory. However, automatic memory management can also be a problem because it takes control away from application developers.

When V8 needs to allocate memory to an object in a Node.js application, it usually does so by using a pointer to assign an object or variable to a new piece of memory. Eventually, the available memory starts to run out, at which point V8 begins a process of garbage collection to free up memory.

The V8 garbage collector must identify which regions of memory can be reallocated without affecting the performance of the application. It follows pointers to find out which objects are currently live and which can not be reached by the existing set of pointers. Memory that contains unreachable objects is called a “dead” region, and can safely be reallocated.

What Causes Memory Leaks?

To avoid memory leaks, V8 needs to know the exact locations of all objects in memory. If objects are identified incorrectly as pointers, memory leaks can result. Memory leaks can occur when the actual memory used by an object extends beyond the memory that has been allocated to that object.

Another potential cause of memory leaks is a callback function, a Javascript function that can be passed as an argument to other code. The V8 garbage collector does not always know whether it is safe to remove objects used in a callback function that can be called multiple times, leading to memory not being reallocated when it should. Callback functions are examples of closures, functions that have access to the variables of the function that called it. Closures are a well-known cause of memory problems because the variables must be left in the memory for as long as closure functions could need to use them.

Node.js memory leaks most commonly occur in objects that contain large collections of data, such as arrays or hashmaps. However, Node.js memory leaks can also occur in other object types, such as strings.

Sometimes, the cause of a memory leak in a Node.js application is not in the application’s code, but rather in an upstream code that the application depends on for its execution. This makes Node.js memory leaks difficult to diagnose. If you have been staring at your Javascript code for ages, and you are convinced that it is not causing your java memory leak, then consider looking at the upstream code for the source of your problem.

In summary, you should consider the following potential causes of a Node.js memory leak:

  • Callbacks and closures
  • Leaky constructors of classes you have defined
  • Collection objects, such as arrays and hashmaps
  • Upstream code

What Are the Symptoms of a Memory Leak in a Node.js Application?

Memory leaks can affect the performance of Node.js applications, causing them to run slowly, malfunction or freeze up completely. Many leaks begin very slowly, but over time, they can cause the app to slow down as V8 needs to spend more and more time in garbage collection mode to manage the app’s memory uses.

Memory leaks can also cause odd behavior in a Node.js application. For example, you might notice that your application stops being able to open new database connections. This occurs because the leaking code is hanging onto references to the resources that the app needs to open new connections. You may find that you need to restart the app to restore its performance.

Not all memory leaks cause noticeable symptoms during the development or testing phase. However, it is still worth trying to find memory leaks in your application so you can eliminate them. As your app becomes more popular, the greater the number of users that will make demands from it. Even a relatively slow memory leak can damage the performance of an app, slowing it down and preventing it from offering the best possible performance to your users. If you let a memory leak continue unchecked, your app will eventually crash, which can be extremely embarrassing and damaging to your reputation among your customers and the online community at large.

Making more RAM available for your app or restarting your service every time its memory usage starts to get out of hand can prevent a crisis. However, it is better to track down the problem and fix it, so your app runs efficiently and consistently without the constant need for you to monitor it or worry about it breaking down.

How Do You Diagnose a Memory Leak in a Node.js Application?

Many metrics can diagnose memory leaks in a Node.js application. These include heap usage, heap growth and frequency of garbage collection. You can also use tools to help you detect leaks in Node.js applications, some of which are listed in this summary from Mozilla.

A heap profiler is an extremely useful tool for diagnosing memory leaks in Node applications. A non-leaky memory profile shows memory usage increasing in response to incoming requests. After the requests are dealt with, you should see memory usage decrease as objects that are no longer needed are destroyed, and the used memory is released.

A leaky memory profile looks very different from a normal profile. Memory usage increases over time. The longer the application is open, the less memory is available, until you reach the point where you must restart the app to reduce the memory usage to normal and restore the app’s performance. Note that memory leaks can be rapid, in which case they are quick and easy to notice in the heap memory usage, or slow, in which case memory usage increases much more gradually.

Taking a look at the instance counts of various types of objects can help to pin down the source of the memory leak in a Node.js application. If the instance count of a particular type of object grows strongly and does not come back down, then this type of object is likely to be the source of the memory leak. Usually, you would expect instance counts of each object to come back down after each garbage collection cycle, so look for objects that don’t have this behavior.

Analyzing these metrics can help you work out which part of your code is causing the problem so you can track down the bug and fix it as quickly as possible. For example, if you have written a leaky class called myClass, then you are likely to see objects of type myClass using up most of the memory in the heap stack. You can then check your definition of the myClass class to look for code that allows a memory leak to occur.

Also, pay attention to the frequency with which the V8 engine runs the garbage collection. When this happens, you might notice your app slowing down. The leakier your code, the more often V8 needs to perform garbage collection.

Dealing With Node.js Memory Leaks

It is important to diagnose and control memory leaks while developing your Node application. Be aware that your application might be experiencing memory leaks from more than one source, so when you think you have eliminated a memory leak, repeat your tests to make sure that the leaks are gone.

Although diagnosing and eliminating memory leaks takes some time, it is worth making the effort to wipe out leaks while developing your app. If you try to build on a leaky code to add more functionality to your app later on, you may find that its performance becomes too detrimental for the app to be useful to your users.

By learning to deal with Node.js memory leaks, you can train yourself to build better apps that work well. You’ll learn to consider the possibility of memory leaks when writing new code. For example, when creating a new class, you need to watch out for constructors that create large objects that don’t get deleted when they should be. You can help to prevent Node.js memory leaks by always considering how and when objects will be created and destroyed while you write your code.

Omed Habib
Omed Habib is a Director of Product Marketing at AppDynamics. He originally joined AppDynamics as a Principal Product Manager to lead the development of their world-class PHP, Node.js and Python APM agents. An engineer at heart, Omed fell in love with web-scale architecture while directing technology throughout his career. He spends his time exploring new ways to help some of the largest software deployments in the world meet their performance needs.

Thank you! Your submission has been received!

Oops! Something went wrong while submitting the form