The 8 symptoms that surface them
The many thousands of java.lang.OutOfMemoryErrors that I’ve met during my career all bear one of the below eight symptoms.
This post from haritibcoblog explains what causes a particular error to be thrown, offers code examples that can cause such errors, and gives you solution guidelines for a fix.
The content is all based on my own experience.
- java.lang.OutOfMemoryError:Java heap space
- java.lang.OutOfMemoryError:GC overhead limit exceeded
- java.lang.OutOfMemoryError:Permgen space
- java.lang.OutOfMemoryError:Unable to create new native thread
- java.lang.OutOfMemoryError:Out of swap space?
- java.lang.OutOfMemoryError:Requested array size exceeds VM limit
- Out of memory:Kill process or sacrifice child
java.lang.OutOfMemoryError:Java heap space
Java applications are only allowed to use a limited amount of memory. This limit is specified during application startup. To make things more complex, Java memory is separated into two different regions. These regions are called Heap space and Permgen (for Permanent Generation):
The size of those regions is set during the Java Virtual Machine (JVM) launch and can be customized by specifying JVM parameters -Xmx and -XX:MaxPermSize. If you do not explicitly set the sizes, platform-specific defaults will be used.
The java.lang.OutOfMemoryError: Java heap space error will be triggered when the application attempts to add more data into the heap space area, but there is not enough room for it.
Note that there might be plenty of physical memory available, but the java.lang.OutOfMemoryError: Java heap space error is thrown whenever the JVM reaches the heap size limit.
What is causing it?
There most common reason for the java.lang.OutOfMemoryError: Java heap space error is simple – you try to fit an XXL application into an S-sized Java heap space. That is – the application just requires more Java heap space than available to it to operate normally. Other causes for this OutOfMemoryError message are more complex and are caused by a programming error:
- Spikes in usage/data volume. The application was designed to handle a certain amount of users or a certain amount of data. When the number of users or the volume of data suddenly spikes and crosses that expected threshold, the operation which functioned normally before the spike ceases to operate and triggers the lang.OutOfMemoryError: Java heap spaceerror.
Memory leaks. A particular type of programming error will lead your application to constantly consume more memory. Every time the leaking functionality of the application is used it leaves some objects behind into the Java heap space. Over time the leaked objects consume all of the available Java heap space and trigger the already familiar java.lang.OutOfMemoryError: Java heap space error.
java.lang.OutOfMemoryError:GC overhead limit exceeded
Java runtime environment contains a built-in Garbage Collection (GC) process. In many other programming languages, the developers need to manually allocate and free memory regions so that the freed memory can be reused.
Java applications on the other hand only need to allocate memory. Whenever a particular space in memory is no longer used, a separate process called Garbage Collection clears the memory for them. How the GC detects that a particular part of memory is explained in more detail in the Garbage Collection Handbook, but you can trust the GC to do its job well.
The java.lang.OutOfMemoryError: GC overhead limit exceeded error is displayed when your application has exhausted pretty much all the available memory and GC has repeatedly failed to clean it.
What is causing it?
The java.lang.OutOfMemoryError: GC overhead limit exceeded error is the JVM’s way of signalling that your application spends too much time doing garbage collection with too little result. By default the JVM is configured to throw this error if it spends more than 98% of the total time doing GC and when after the GC only less than 2% of the heap is recovered.
What would happen if this GC overhead limit would not exist? Note that the java.lang.OutOfMemoryError: GC overhead limit exceeded error is only thrown when 2% of the memory is freed after several GC cycles. This means that the small amount of heap the GC is able to clean will likely be quickly filled again, forcing the GC to restart the cleaning process again. This forms a vicious cycle where the CPU is 100% busy with GC and no actual work can be done. End users of the application face extreme slowdowns – operations which normally complete in milliseconds take minutes to finish.
So the “java.lang.OutOfMemoryError: GC overhead limit exceeded” message is a pretty nice example of a fail fast principle in action.
Java applications are only allowed to use a limited amount of memory. The exact amount of memory your particular application can use is specified during application startup. To make things more complex, Java memory is separated into different regions which can be seen in the following figure:
The size of all those regions, including the permgen area, is set during the JVM launch. If you do not set the sizes yourself, platform-specific defaults will be used.
The java.lang.OutOfMemoryError: PermGen space message indicates that the Permanent Generation’s area in memory is exhausted.
What is causing it?
To understand the cause for the java.lang.OutOfMemoryError: PermGen space, we would need to understand what this specific memory area is used for.
For practical purposes, the permanent generation consists mostly of class declarations loaded and stored into PermGen. This includes the name and fields of the class, methods with the method bytecode, constant pool information, object arrays and type arrays associated with a class and Just In Time compiler optimizations.
From the above definition you can deduce that the PermGen size requirements depend both on the number of classes loaded as well as the size of such class declarations. Therefore we can say that the main cause for the java.lang.OutOfMemoryError: PermGen space is that either too many classes or too big classes are loaded to the permanent generation.
Java applications are allowed to use only a limited amount of memory. The exact amount of memory your particular application can use is specified during application startup. To make things more complex, Java memory is separated into different regions, as seen in the following figure:
The size of all those regions, including the metaspace area, can be specified during the JVM launch. If you do not determine the sizes yourself, platform-specific defaults will be used.
The java.lang.OutOfMemoryError: Metaspace message indicates that the Metaspace area in memory is exhausted.
What is causing it?
If you are not a newcomer to the Java landscape, you might be familiar with another concept in Java memory management called PermGen. Starting from Java 8, the memory model in Java was significantly changed. A new memory area called Metaspace was introduced and Permgen was removed. This change was made due to variety of reasons, including but not limited to:
- The required size of permgen was hard to predict. It resulted in either under-provisioning triggering lang.OutOfMemoryError: Permgen sizeerrors or over-provisioning resulting in wasted resources.
- GC performanceimprovements, enabling concurrent class data de-allocation without GC pauses and specific iterators on metadata
- Support for further optimizations such as G1concurrent class unloading.
So if you were familiar with PermGen then all you need to know as background is that – whatever was in PermGen before Java 8 (name and fields of the class, methods of a class with the bytecode of the methods, constant pool, JIT optimizations etc) – is now located in Metaspace.
As you can see, Metaspace size requirements depend both upon the number of classes loaded as well as the size of such class declarations. So it is easy to see the main cause for the java.lang.OutOfMemoryError: Metaspace is: either too many classes or too big classes being loaded to the Metaspace.
java.lang.OutOfMemoryError:Unable to create new native thread
Java applications are multi-threaded by nature. What this means is that the programs written in Java can do several things (seemingly) at once. For example – even on machines with just one processor – while you drag content from one window to another, the movie played in the background does not stop just because you carry out several operations at once.
A way to think about threads is to think of them as workers to whom you can submit tasks to carry out. If you had only one worker, he or she could only carry out one task at the time. But when you have a dozen workers at your disposal they can simultaneously fulfill several of your commands.
Now, as with workers in physical world, threads within the JVM need some elbow room to carry out the work they are summoned to deal with. When there are more threads than there is room in memory we have built a foundation for a problem:
The message java.lang.OutOfMemoryError: Unable to create new native thread means that the Java application has hit the limit of how many Threads it can launch.
What is causing it?
You have a chance to face the java.lang.OutOfMemoryError: Unable to create new native thread whenever the JVM asks for a new thread from the OS. Whenever the underlying OS cannot allocate a new native thread, this OutOfMemoryError will be thrown. The exact limit for native threads is very platform-dependent thus we recommend to find out those limits by running a test similar to the below example. But, in general, the situation causing java.lang.OutOfMemoryError: Unable to create new native thread goes through the following phases:
- A new Java thread is requested by an application running inside the JVM
- JVM native code proxies the request to create a new native thread to the OS
- The OS tries to create a new native thread which requires memory to be allocated to the thread
- The OS will refuse native memory allocation either because the 32-bit Java process size has depleted its memory address space – e.g. (2-4) GB process size limit has been hit – or the virtual memory of the OS has been fully depleted
- The lang.OutOfMemoryError: Unable to create new native threaderror is thrown.
java.lang.OutOfMemoryError:Out of swap space?
Java applications are given limited amount of memory during the startup. This limit is specified via the -Xmx and other similar startup parameters. In situations where the total memory requested by the JVM is larger than the available physical memory, operating system starts swapping out the content from memory to hard drive.
The java.lang.OutOfMemoryError: Out of swap space? error indicates that the swap space is also exhausted and the new attempted allocation fails due to the lack of both physical memory and swap space.
What is causing it?
The java.lang.OutOfmemoryError: Out of swap space? is thrown by JVM when an allocation request for bytes from the native heap fails and the native heap is close to exhaustion. The message indicates the size (in bytes) of the allocation which failed and the reason for the memory request.
The problem occurs in situations where the Java processes have started swapping, which, recalling that Java is a garbage collected language is already not a good situation. Modern GC algorithms do a good job, but when faced with latency issues caused by swapping, the GC pauses tend to increase to levels not tolerable by most applications.
java.lang.OutOfMemoryError: Out of swap space? is often caused by operating system level issues, such as:
- The operating system is configured with insufficient swap space.
- Another process on the system is consuming all memory resources.
It is also possible that the application fails due to a native leak, for example, if application or library code continuously allocates memory but does not release it to the operating system.
java.lang.OutOfMemoryError:Requested array size exceeds VM limit
Java has got a limit on the maximum array size your program can allocate. The exact limit is platform-specific but is generally somewhere between 1 and 2.1 billion elements.
When you face the java.lang.OutOfMemoryError: Requested array size exceeds VM limit, this means that the application that crashes with the error is trying to allocate an array larger than the Java Virtual Machine can support.
What is causing it?
The error is thrown by the native code within the JVM. It happens before allocating memory for an array when the JVM performs a platform-specific check: whether the allocated data structure is addressable in this platform. This error is less common than you might initially think.
The reason you only seldom face this error is that Java arrays are indexed by int. The maximum positive int in Java is 2^31 – 1 = 2,147,483,647. And the platform-specific limits can be really close to this number – for example on my 64bit MB Pro on Java 1.7 I can happily initialize arrays with up to 2,147,483,645 or Integer.MAX_VALUE-2 elements.
Increasing the length of the array by one to Integer.MAX_VALUE-1 results in the familiar OutOfMemoryError:
Exception in thread “main” java.lang.OutOfMemoryError: Requested array size exceeds VM limit
But the limit might not be that high – on 32-bit Linux with OpenJDK 6, you will hit the “java.lang.OutOfMemoryError: Requested array size exceeds VM limit” already when allocating an array with ~1.1 billion elements. To understand the limits of your specific environments run the small test program described in the next chapter.
Out of memory:Kill process or sacrifice child
In order to understand this error, we need to recoup the operating system basics. As you know, operating systems are built on the concept of processes. Those processes are shepherded by several kernel jobs, one of which, named “Out of memory killer” is of interest to us in this particular case.
This kernel job can annihilate your processes under extremely low memory conditions. When such a condition is detected, the Out of memory killer is activated and picks a process to kill. The target is picked using a set of heuristics scoring all processes and selecting the one with the worst score to kill. The Out of memory: Kill process or sacrifice child is thus different from other errors covered in our OOM handbook as it is not triggered nor proxied by the JVM but is a safety net built into the operating system kernels.
The Out of memory: kill process or sacrifice child error is generated when the available virtual memory (including swap) is consumed to the extent where the overall operating system stability is put to risk. In such case the Out of memory killer picks the rogue process and kills it.
What is causing it?
By default, Linux kernels allow processes to request more memory than currently available in the system. This makes all the sense in the world, considering that most of the processes never actually use all of the memory they allocate. The easiest comparison to this approach would be the broadband operators. They sell all the consumers a 100Mbit download promise, far exceeding the actual bandwidth present in their network. The bet is again on the fact that the users will not simultaneously all use their allocated download limit. Thus one 10Gbit link can successfully serve way more than the 100 users our simple math would permit.
A side effect of such an approach is visible in case some of your programs are on the path of depleting the system’s memory. This can lead to extremely low memory conditions, where no pages can be allocated to process. You might have faced such situation, where not even a root account cannot kill the offending task. To prevent such situations, the killer activates, and identifies the rogue process to be the killed.
You can read more about fine-tuning the behaviour of “Out of memory killer” in this article from RedHat documentation.
Now that we have the context, how can you know what triggered the “killer” and woke you up at 5AM? One common trigger for the activation is hidden in the operating system configuration. When you check the configuration in /proc/sys/vm/overcommit_memory, you have the first hint – the value specified here indicates whether all malloc() calls are allowed to succeed. Note that the path to the parameter in the proc file system varies depending on the system affected by the change.
Overcommitting configuration allows to allocate more and more memory for this rogue process which can eventually trigger the “Out of memory killer” to do exactly what it is meant to do.