|By Jinwoo Hwang||
|May 8, 2010 06:45 AM EDT||
When we encounter a java.lang.OutOfMemoryError, we often find that Java heap dumps, along with other artifacts, are generated by the Java Virtual Machine. If you feel like jumping right into a Java heap dump when you get a java.lang.OutOfMemoryError, don't worry, it's a normal thought. You may be able to discover something serendipitously, but it's not always the best idea to analyze Java heap dumps, depending on the situation you are facing. We first need to investigate the root cause of the java.lang.OutOfMemoryError.
Only after the root cause is identified can we decide whether or not to analyze Java heap dumps. What is a java.lang.OutOfMemoryError? Why in the world does it occur? Let's find out.
What Is a java.lang.OutOfMemoryError?
A java.lang.OutOfMemoryError is a subclass of java.lang.VirtualMachineError that is thrown when the Java Virtual Machine is broken or has run out of resources that are necessary to continue the operation of the Java Virtual Machine. Obviously, memory is the exhausted resource for a java.lang.OutOfMemoryError, which is thrown when the Java Virtual Machine cannot allocate an object due to memory constraints. Unfortunately, the Java specification of java.lang.OutOfMemoryError does not elaborate further on what kind of memory it's talking about.
There are six different types of runtime data areas, or memory areas, in the Java Virtual Machine (see Figure 1).
- Program Counter Register
- Java Virtual Machine Stack
- Method Area
- Runtime Constant Pool
- Native Method Stack
The Program Counter Register, also known as the pc register, stores the address of the Java byte code instruction that is currently being executed (just like the processor register in your central processing unit of the device from which you are reading or printing this article). You will not see a java.lang.OutOfMemoryError from the pc register since a program counter is not conventionally considered as a memory.
Java Virtual Machine Stacks contain frames where data, return values, and partial execution results are stored. Java Virtual Machine Stacks can be expanded during runtime. If there's not enough memory for the expansion of an existing Java Virtual Machine stack, or for the creation of a new Java Virtual Machine stack for a new thread, the Java Virtual Machine will throw a java.lang.OutOfMemoryError.
The Heap is where instances of Java classes and arrays are allocated. A java.lang.OutOfMemoryError will be thrown when there is not enough memory available for instances of Java classes or arrays.
The Method Area stores class-related information, the runtime constant pool, for instances, the code for methods and constructors, and field/method data. If there's not enough memory in the method area, you will encounter java.lang.OutOfMemoryError.
The Runtime Constant Pool contains constants such as field references and literals. A java.lang.OutOfMemoryError will be thrown when not enough memory is available for the construction of the runtime constant pool area.
Native Memory Stacks store conventional stacks, also known as C stacks, to support native methods that are written in a non-Java language such as C/C++. Native memory stacks can be expanded during runtime. If there's not enough memory for the expansion of an existing native memory stack or for the creation of a new native memory stack for a new thread, you would see a java.lang.OutOfMemoryError.
You may have seen a java.lang.StackOverflowError, which is completely different from a java.lang.OutOfMemoryError. A java.lang.StackOverflowError is thrown when native memory stacks or Java Virtual Machine stacks need more memory than is configured. In most IBM Java Virtual Machine implementations, the -Xmso command-line option controls the stack size for operation system threads or native thread, and the -Xss command-line option controls the stack size for Java threads. In some implementations, such as Sun Microsystems HotSpot Java Virtual Machine, the Java methods share stack frames with C/C++ native code. The maximum stack size for a thread can be configured with the -Xss Java command-line option. The default sizes of these options vary by platform and implementation, but are usually between 256 Kbytes-1024 Kbytes. Please refer to the documentation of your Java virtual machine for more specific information. We will cover more about java.lang.StackOverflowError in a separate article.
Now that we understand which memory areas could cause a java.lang.OutOfMemoryError, let's take a look at actual error messages. What does a java.lang.OutOfMemoryError look like and how can I address each symptom? Have you ever seen a java.lang.OutOfMemoryError similar to the following?
java.lang.OutOfMemoryError: Requested array size exceeds VM limit
This error message indicates that there is a memory request for an array but that's too large for a predefined limit of a virtual machine. What do we do if we encounter this kind of java.lang.OutOfMemoryError? We need to check the source code to make sure that there's no huge array created dynamically or statically. Fortunately, latest virtual machines usually do not have this limit.
java.lang.OutOfMemoryError: PermGen space
You will see an OutOfMemoryError when the Permanent Generation area of the Java heap is full, like the above message.
On some Java Virtual Machines, such as Sun Microsystems' HotSpot Java Virtual Machine, a dedicated memory area called permanent generation (or permanent region) stores objects that describe classes and methods. We can visualize the usage of a permanent generation with the IBM Pattern Modeling and Analysis Tool for the Java Garbage Collector.
In Figure 2 we enabled the "Max Perm" button and the "Used Tenured" button to visualize permanent generation usage and its maximum size. We can see that the used amount of permanent generation reaches its maximum limit. That's why we're getting the java.lang.OutOfMemoryError: PermGen space message. If there's no memory leak, we can just use the -XX:MaxPermSize command-line option to increase the maximum limit of the permanent generation. For example,
will set the maximum size of the permanent generation to 128 Mbytes.
So far we've seen a Java.lang.OutOfMemoryError due to exhaustion in the Java heap or an area in the Java heap such as permanent generation. Surprisingly, a Java.lang.OutOfMemoryError can be thrown when the Java Virtual Machine cannot find any more memory in the native memory as well as in the Java heap. How can we tell whether it's caused by the Java heap or native memory?
In the following message, there's no information in the message whether java.lang.OutOfMemoryError is caused by the Java heap or native memory:
JVMDUMP013I Processed dump event "systhrow", detail "java/lang/OutOfMemoryError".
In the following case, the Java virtual machine is kind enough to tell us that there's native memory exhaustion. In the message, the Java virtual machine says "allocateMemory failed" which means a native memory allocation failed:
java.lang.OutOfMemoryError: JVMCI046: allocateMemory failed
In the following message, there's no clue as to whether it's native memory or a Java heap. Fortunately we have a line number, 20, and the source code file name, HeapExhaustionSimulator.java. This might be Java heap related.
JVMDG274: Dump Handler has Processed OutOfMemory.
Exception in thread "main" java.lang.OutOfMemoryError
In the following message, there's no clue whether it's native memory or a Java heap. But "sun.misc.Unsafe.allocateMemory(Native Method)" indicates that it might be native memory related.
Exception in thread "main" java.lang.OutOfMemoryError
at sun.misc.Unsafe.allocateMemory(Native Method)
In the following message, the Java Virtual Machine indicates that the Java heap space is related to the java.lang.OutOfMemoryError.
java.lang.OutOfMemoryError: Java heap space
Dumping heap to java_pid6280.hprof ...
Heap dump file created [50549348 bytes in 1.444 secs]
You may have seen a java.lang.OutOfMemoryError similar to the following:
java.lang.OutOfMemoryError: requested NNN bytes for MMMM. Out of swap space?
Literally you could check the operating system configuration for swap space. It seems that the Java Virtual Machine is not sure if the swap space is the root cause of the problem (?).We can check whether this Java Virtual Machine is consuming too much native memory .We also need to make sure there's enough memory for this JVM and no other processes are consuming most of memory resource. The last thing we can try is to find any known defects related to the module, MMMM.
java.lang.OutOfMemoryError: unable to create new native thread
This kind of message is seen when you have an excessive number of threads or if the native memory is exhausted and a thread is attempting to be created.
What Is a Java Heap Dump?
We've learned that a Java heap is a runtime data area where all class instances and arrays are allocated and shared among all Java Virtual Machine threads during execution of the JVM. A Java heap dump is a snapshot of a Java heap at a specific time. It's like taking a picture of a busy warehouse at a given time. If we look at the picture, we can identify what items were available at that time. Some items may be shipped to Canada a few minutes later, but you can see them in the picture because they were there at the time of the snapshot.
Because the Java specification does not mention the Java heap dump, there are different forms of Java heap dump implementations from different Java Virtual Machines. The IBM Java heap dump provides information mostly about the Java heap.
The Sun Microsystems hprof Java heap dump provides information about the Java Virtual Machine stacks, the runtime constant pool as well as the Java heap.
How Can I Generate Java Heap Dumps?
A Java heap dump is usually automatically generated by the Java Virtual Machine, but you can also force Java heap dump generation. On most IBM Java Virtual Machines, Java heap dumps are generated automatically when the Java heap becomes exhausted. On most Sun Microsystems JVMs, you need to configure the virtual machine to generate Java heap dumps. If you want to generate a Java heap dump when a java.lang.OutOfMemoryError occurs, you need to set the -XX:+HeapDumpOnOutOfMemoryError command-line option on certain releases of Sun's JVM. You could also use a HPROF profiler by using the -agentlib:hprof=heap=dump command-line option. You could also use jmap if your Sun JVM provides the utility. For example, jmap -dump 1234 will generate the Java heap dump from the process whose identifier is 1234. You could utilize JConsole by calling the HotSpotDiagnostic MBean and the dumpHeap operation if it's available from your Sun JVM.
If you want to generate Java heap dumps for Java virtual machine crashes (an unexpected termination of process) or user signals on IBM JVMs, you can set the environment variable IBM_HEAPDUMP or IBM_HEAP_DUMP to TRUE. For example, you can send the IBM Java virtual machine the signal SIGQUIT for the Linux operating systems and AIX operating systems or SIGINT(Control-Break key combination) for Windows to generate Java heap dumps. The IBM JVM provides an API, com.ibm.jvm.Dump.HeapDump(), that you can invoke from application code to generate Java heap dumps programmatically.
Please refer to the documentation of your JVM for detailed information since these options vary by platform and implementation.
Where Can I Find Java Heap Dumps?
You can find Java heap dumps in the current working directory of the Java Virtual Machine process, unless you specify a different location. You can specify the location with the environment variable IBM_HEAPDUMPDIR or _CEE_DMPTARG on IBM JVMs. If there's not enough space available for Java heap dumps or the JVM cannot acquire write-permission in the location, Java heap dumps are generated to the operating system's temporary directory on the IBM JVM. Please refer to your operating system manual for the location of the system's temporary directory and its configuration.
What Do Java Heap Dumps Look Like and How Can I read Them?
Nowadays, the majority of Java heap dump formats are generated in binary. Thus you might want to use a tool unless your brain can interpret hexadecimal codes without any headaches.
The IBM HeapAnalyzer is one of the most popular Java heap analysis tools. It can analyze all Java heap dump formats provided by Sun, HP and most of the Java heap dump formats provided by IBM. It's powered by object query engines and patent-pending heuristic analysis engines. I've been developing the IBM HeapAnalyzer from scratch since 2003, spending my vacation, weekends and weeknights on it. The IBM HeapAnalyzer was so successful that IBM decided to make the IBM HeapAnalyzer an official IBM software product and bundle it with existing products. So I donated all the source code of IBM HeapAnalyzer to IBM to run it on an Eclipse-based interface. Now the IBM HeapAnalyzer has a daughter, MDD4J, which always reminds me of my late daughter lost while I was working on the MDD4J project. The IBM HeapAnalyzer has been the top technology at IBM alphaWorks for six consecutive years since its inception as of March 2009.
Whether your Java heap dump is in binary or text/ascii format, the heap dump contains information about all the live objects that are on the Java heap such as address, object size, and referenced addresses. Let's take a look at the text/ascii format Java heap dumps since binary heap dumps have similar information but are in hexadecimal format to save disk space.
Internet of Things (IoT) will be a hybrid ecosystem of diverse devices and sensors collaborating with operational and enterprise systems to create the next big application. In their session at @ThingsExpo, Bramh Gupta, founder and CEO of robomq.io, and Fred Yatzeck, principal architect leading product development at robomq.io, discussed how choosing the right middleware and integration strategy from the get-go will enable IoT solution developers to adapt and grow with the industry, while at the same time reduce Time to Market (TTM) by using plug and play capabilities offered by a robust IoT ...
Oct. 9, 2015 02:00 AM EDT Reads: 2,191
Today’s connected world is moving from devices towards things, what this means is that by using increasingly low cost sensors embedded in devices we can create many new use cases. These span across use cases in cities, vehicles, home, offices, factories, retail environments, worksites, health, logistics, and health. These use cases rely on ubiquitous connectivity and generate massive amounts of data at scale. These technologies enable new business opportunities, ways to optimize and automate, along with new ways to engage with users.
Oct. 9, 2015 02:00 AM EDT Reads: 160
The buzz continues for cloud, data analytics and the Internet of Things (IoT) and their collective impact across all industries. But a new conversation is emerging - how do companies use industry disruption and technology enablers to lead in markets undergoing change, uncertainty and ambiguity? Organizations of all sizes need to evolve and transform, often under massive pressure, as industry lines blur and merge and traditional business models are assaulted and turned upside down. In this new data-driven world, marketplaces reign supreme while interoperability, APIs and applications deliver un...
Oct. 9, 2015 02:00 AM EDT Reads: 278
The Internet of Things (IoT) is growing rapidly by extending current technologies, products and networks. By 2020, Cisco estimates there will be 50 billion connected devices. Gartner has forecast revenues of over $300 billion, just to IoT suppliers. Now is the time to figure out how you’ll make money – not just create innovative products. With hundreds of new products and companies jumping into the IoT fray every month, there’s no shortage of innovation. Despite this, McKinsey/VisionMobile data shows "less than 10 percent of IoT developers are making enough to support a reasonably sized team....
Oct. 9, 2015 02:00 AM EDT Reads: 202
“In the past year we've seen a lot of stabilization of WebRTC. You can now use it in production with a far greater degree of certainty. A lot of the real developments in the past year have been in things like the data channel, which will enable a whole new type of application," explained Peter Dunkley, Technical Director at Acision, in this SYS-CON.tv interview at @ThingsExpo, held Nov 4–6, 2014, at the Santa Clara Convention Center in Santa Clara, CA.
Oct. 9, 2015 01:45 AM EDT Reads: 7,008
Through WebRTC, audio and video communications are being embedded more easily than ever into applications, helping carriers, enterprises and independent software vendors deliver greater functionality to their end users. With today’s business world increasingly focused on outcomes, users’ growing calls for ease of use, and businesses craving smarter, tighter integration, what’s the next step in delivering a richer, more immersive experience? That richer, more fully integrated experience comes about through a Communications Platform as a Service which allows for messaging, screen sharing, video...
Oct. 9, 2015 12:00 AM EDT Reads: 1,130
SYS-CON Events announced today that Dyn, the worldwide leader in Internet Performance, will exhibit at SYS-CON's 17th International Cloud Expo®, which will take place on November 3-5, 2015, at the Santa Clara Convention Center in Santa Clara, CA. Dyn is a cloud-based Internet Performance company. Dyn helps companies monitor, control, and optimize online infrastructure for an exceptional end-user experience. Through a world-class network and unrivaled, objective intelligence into Internet conditions, Dyn ensures traffic gets delivered faster, safer, and more reliably than ever.
Oct. 8, 2015 10:00 PM EDT Reads: 590
There are so many tools and techniques for data analytics that even for a data scientist the choices, possible systems, and even the types of data can be daunting. In his session at @ThingsExpo, Chris Harrold, Global CTO for Big Data Solutions for EMC Corporation, will show how to perform a simple, but meaningful analysis of social sentiment data using freely available tools that take only minutes to download and install. Participants will get the download information, scripts, and complete end-to-end walkthrough of the analysis from start to finish. Participants will also be given the pract...
Oct. 8, 2015 09:15 PM EDT Reads: 285
The IoT market is on track to hit $7.1 trillion in 2020. The reality is that only a handful of companies are ready for this massive demand. There are a lot of barriers, paint points, traps, and hidden roadblocks. How can we deal with these issues and challenges? The paradigm has changed. Old-style ad-hoc trial-and-error ways will certainly lead you to the dead end. What is mandatory is an overarching and adaptive approach to effectively handle the rapid changes and exponential growth.
Oct. 8, 2015 09:00 PM EDT Reads: 116
Mobile messaging has been a popular communication channel for more than 20 years. Finnish engineer Matti Makkonen invented the idea for SMS (Short Message Service) in 1984, making his vision a reality on December 3, 1992 by sending the first message ("Happy Christmas") from a PC to a cell phone. Since then, the technology has evolved immensely, from both a technology standpoint, and in our everyday uses for it. Originally used for person-to-person (P2P) communication, i.e., Sally sends a text message to Betty – mobile messaging now offers tremendous value to businesses for customer and empl...
Oct. 8, 2015 05:30 PM EDT Reads: 232
Can call centers hang up the phones for good? Intuitive Solutions did. WebRTC enabled this contact center provider to eliminate antiquated telephony and desktop phone infrastructure with a pure web-based solution, allowing them to expand beyond brick-and-mortar confines to a home-based agent model. It also ensured scalability and better service for customers, including MUY! Companies, one of the country's largest franchise restaurant companies with 232 Pizza Hut locations. This is one example of WebRTC adoption today, but the potential is limitless when powered by IoT.
Oct. 8, 2015 04:30 PM EDT Reads: 7,471
You have your devices and your data, but what about the rest of your Internet of Things story? Two popular classes of technologies that nicely handle the Big Data analytics for Internet of Things are Apache Hadoop and NoSQL. Hadoop is designed for parallelizing analytical work across many servers and is ideal for the massive data volumes you create with IoT devices. NoSQL databases such as Apache HBase are ideal for storing and retrieving IoT data as “time series data.”
Oct. 8, 2015 02:45 PM EDT Reads: 498
Clearly the way forward is to move to cloud be it bare metal, VMs or containers. One aspect of the current public clouds that is slowing this cloud migration is cloud lock-in. Every cloud vendor is trying to make it very difficult to move out once a customer has chosen their cloud. In his session at 17th Cloud Expo, Naveen Nimmu, CEO of Clouber, Inc., will advocate that making the inter-cloud migration as simple as changing airlines would help the entire industry to quickly adopt the cloud without worrying about any lock-in fears. In fact by having standard APIs for IaaS would help PaaS expl...
Oct. 8, 2015 02:30 PM EDT Reads: 652
NHK, Japan Broadcasting, will feature the upcoming @ThingsExpo Silicon Valley in a special 'Internet of Things' and smart technology documentary that will be filmed on the expo floor between November 3 to 5, 2015, in Santa Clara. NHK is the sole public TV network in Japan equivalent to the BBC in the UK and the largest in Asia with many award-winning science and technology programs. Japanese TV is producing a documentary about IoT and Smart technology and will be covering @ThingsExpo Silicon Valley. The program, to be aired during the peak viewership season of the year, will have a major impac...
Oct. 8, 2015 01:00 PM EDT Reads: 261
SYS-CON Events announced today that ProfitBricks, the provider of painless cloud infrastructure, will exhibit at SYS-CON's 17th International Cloud Expo®, which will take place on November 3–5, 2015, at the Santa Clara Convention Center in Santa Clara, CA. ProfitBricks is the IaaS provider that offers a painless cloud experience for all IT users, with no learning curve. ProfitBricks boasts flexible cloud servers and networking, an integrated Data Center Designer tool for visual control over the cloud and the best price/performance value available. ProfitBricks was named one of the coolest Clo...
Oct. 8, 2015 01:00 PM EDT Reads: 763
Organizations already struggle with the simple collection of data resulting from the proliferation of IoT, lacking the right infrastructure to manage it. They can't only rely on the cloud to collect and utilize this data because many applications still require dedicated infrastructure for security, redundancy, performance, etc. In his session at 17th Cloud Expo, Emil Sayegh, CEO of Codero Hosting, will discuss how in order to resolve the inherent issues, companies need to combine dedicated and cloud solutions through hybrid hosting – a sustainable solution for the data required to manage I...
Oct. 8, 2015 01:00 PM EDT Reads: 477
Apps and devices shouldn't stop working when there's limited or no network connectivity. Learn how to bring data stored in a cloud database to the edge of the network (and back again) whenever an Internet connection is available. In his session at 17th Cloud Expo, Bradley Holt, Developer Advocate at IBM Cloud Data Services, will demonstrate techniques for replicating cloud databases with devices in order to build offline-first mobile or Internet of Things (IoT) apps that can provide a better, faster user experience, both offline and online. The focus of this talk will be on IBM Cloudant, Apa...
Oct. 8, 2015 12:45 PM EDT Reads: 508
WebRTC is about the data channel as much as about video and audio conferencing. However, basically all commercial WebRTC applications have been built with a focus on audio and video. The handling of “data” has been limited to text chat and file download – all other data sharing seems to end with screensharing. What is holding back a more intensive use of peer-to-peer data? In her session at @ThingsExpo, Dr Silvia Pfeiffer, WebRTC Applications Team Lead at National ICT Australia, will look at different existing uses of peer-to-peer data sharing and how it can become useful in a live session to...
Oct. 8, 2015 12:00 PM EDT Reads: 607
As a company adopts a DevOps approach to software development, what are key things that both the Dev and Ops side of the business must keep in mind to ensure effective continuous delivery? In his session at DevOps Summit, Mark Hydar, Head of DevOps, Ericsson TV Platforms, will share best practices and provide helpful tips for Ops teams to adopt an open line of communication with the development side of the house to ensure success between the two sides.
Oct. 8, 2015 12:00 PM EDT Reads: 576
SYS-CON Events announced today that IBM Cloud Data Services has been named “Bronze Sponsor” of SYS-CON's 17th Cloud Expo, which will take place on November 3–5, 2015, at the Santa Clara Convention Center in Santa Clara, CA. IBM Cloud Data Services offers a portfolio of integrated, best-of-breed cloud data services for developers focused on mobile computing and analytics use cases.
Oct. 8, 2015 11:00 AM EDT Reads: 728