How to detect a memory leak?

Hi,
we are running a web application with Tomcat 4.1.27 on HP-UX 11i under HP-JVM 1.4.2_05.
We have major problem, that the application suddenly starts to allocated the whole memory (6GB) and the server stops running because the JVM is busy doing GCs. We don`t get an out of memory exception, but after 300sec or more doing major-gc, only about 100mb memory is freed, which is allocated again very very fast.
I tried -verbosegc and the application was profiled with jprobe on hp-ux environmen tand with optimizeit on windows environment. With the profling tools we didn`t find any memory leaks. We even used verbosegc in production environment, the server runs fine for days spending about 0.09 seks for minor GCs and it allocates about 1gb(256m for Eden) of 6gb. But suddenly the server starts to do minor gcs about 2 to 4 seks and after a few minutes the whole memory is allocated and the vm is busy doing major GCs. In the log-files, which are very detailed, we can`t find any coherency with the suddenly appearing GCs.
Any approach to solve this problem?

I`ve found it:
I have to use -Xeprof with the following option:
"time_on=sigusr1|sigusr2 (SDK 1.4.1 and later)
Specifies which signal will cause profiling to begin (profile data collection). Please be aware that the application or the VM may already be using the sigusr signals for their own purposes; please check the documentation. Specifying a signal and a timeout (see above) at the same time is possible by repeating the time_on option. Only one of the two signals can be declared to use as the signal to start profiling. During the application's run, the specified signal can be delivered to the Java� process multiple times."
But i have no idea how to define a signal which causes the profiling to begin.
What is a sigusr signal? And how to specify this signal?
Are these signals or Java or HP-UX?

Similar Messages

How to define the memory leak in stability test

Hi
Usually, we run our system with 70% CPU load for 72 hours for stability test (on solaris 10), because some plugin of our system is using mtmalloc, and we use prstat to monitor the memory of each plugin. But because of the complex memory usage of our application, we don't know when the application will use the Max memory, and because of the "mtmalloc", the memory showed by "prstat" will not released but continually increased.
So fro the test point of view, there is the risk of memory leak, but actually, there may be no memory leak, so my question is how to define the memory leak in such condition.

kevin wrote:
Thank you for the input and all the info.
isn't java heap the same as memory allocated to the java process in my weblogic
starup script ?The heap is sized by the -Xmx and -Xms parameters you pass on the java
command-line. The permanent generation is separate.
>
let me also download jprobe and try to run it and see what it gives me.
I'd start by running with -verbose:gc. I'd want to know whether you're
running out of heap or permgen space.
-- Rob
kevin
Rob Woollen <[email protected]> wrote:
Unfortunately memory leaks are not fun to track down even with tools.
I'd first suggest determining whether you're running out of space in
the
permanent area (where classes are loaded), or you've exhausted the java
help space.
I'd start by adding -verbose:gc. Look at the gc messages right before
you hit the OutOfMemoryError. If there's plenty of space left, I'd
suspect you're running out of perm space. Search these newsgroups and
the web for MaxPermSize, and you should see plenty of info.
If you're running out of java heap, tools like jprobe and OptimizeIt
are
helpful. If you can tell me a little more about your application and
how you're testing it, I can offer some more tips.
-- Rob
kevin wrote:
Iam new to JAVA and weblogic. I have an application that runs out ofmemory time
and again.
please let me know how to pin point this problem and moreover, howto interpret
or understand that there is a problem. I have downloaded JPROFILE tool,but it
is very confusing to understand what is goin on in this tool.
If somebody can let me know how to interpret and understand the memoryleak, that
will be great !!!
thank you.

How to deal with Memory Leaks, that are caused by Binding

Hi, I recently noticed (huge?) memory leaks in my application and suspect bindings to be the cause of all the evil.
I made a little test case, which confirms my suspicion:
import javafx.application.Application;
import javafx.beans.property.SimpleStringProperty;
import javafx.beans.property.StringProperty;
import javafx.scene.Scene;
import javafx.scene.control.Button;
import javafx.scene.layout.VBox;
import javafx.stage.Stage;
public class TestAppMemoryLeak extends Application {
    public static void main(String[] args) {
        launch(args);
    @Override
    public void start(Stage stage) throws Exception {
        VBox root = new VBox();
        Button button = null;
        for (int i = 0; i < 100000; i++) {
            button = new Button();
            button.textProperty().bind(text);
            button.textProperty().unbind(); // if you don't call this, you can notice the increased memory of the java process.
        root.getChildren().add(button);
        Scene scene = new Scene(root);
        stage.setScene(scene);
        stage.show();
    private StringProperty text = new SimpleStringProperty("test");
}Now the problem is, HOW can I know, when a variable is no longer needed or overwritten by a new instance.
Just an example:
I have a ListView with a Cell Factory. In the updateItem method, I add a ContextMenu. The textProperty of each MenuItem is bound to a kind of global property, like in the example above. I have to do it in the updateItem method, since the ContextMenu differs depending on the item.
So every time the updateItem method is called a new ContextMenu is created, which binds some properties, but the old context menus remain in memory.
I guess there could be many more example.
How can I deal with it?

I've dealt with this situation and created a Jira issue for it, but I also have a work-around that is a bit unwieldy but works. I'll share it with you.
The bug that deals with this (in part atleast): http://javafx-jira.kenai.com/browse/RT-20616
The solution is to use weak invalidation listeners, however they are a bit of a pain to use as you cannot do it with something simplistic as "bindWeakly"... and you need to keep a reference around to the wrapped listener otherwise it will just get garbage collected immediately (as it is only weakly referenced). Some very odd bugs can surface if weak listeners disappear randomly because you forgot to reference them :)
Anyway, see this code below, it shows you some code that is called from a TreeCell's updateItem method (I've wrapped it in some more layers in my program, but it is essentially the same as an updateItem method):
public class EpisodeCell extends DuoLineCell implements MediaNodeCell {
private final WeakBinder binder = new WeakBinder();
@Override
public void configureCell(MediaNode mediaNode) {
    MediaItem item = mediaNode.getMediaItem();
    StringBinding episodeRange = MapBindings.selectString(mediaNode.dataMapProperty(), Episode.class, "episodeRange");
    binder.unbindAll();
    binder.bind(titleProperty(), MapBindings.selectString(mediaNode.dataMapProperty(), Media.class, "title"));
    binder.bind(ratingProperty(), MapBindings.selectDouble(mediaNode.dataMapProperty(), Media.class, "rating").divide(10));
    binder.bind(extraInfoProperty(), Bindings.when(episodeRange.isNull()).then(new SimpleStringProperty("Special")).otherwise(episodeRange));
    binder.bind(viewedProperty(), item.viewedProperty());
    subtitleProperty().set("");
}This code makes use of a class called WeakBinder -- it is a helper class that can make Weak bindings and can keep track of them. When you call unbindAll() on it, all of the bindings it created before are released immediately (although they will also disappear when the Cell itself is garbage collected, which is possible because it only makes weak references).
I've tested this extensively and it solves the problem of Cells keeping references to objects with much longer life cycles (in my case, the MediaNode passed in has a longer lifecycle than the cells and so it is important to bind weakly to it). Before this would create huge memory leaks (crashing my program within a minute if you kept refreshing the Tree)... now it survives hours atleast and the Heap usage stays in a fixed range which means it is correctly able to collect all garbage).
The code for WeakBinder is below (you can consider it public domain, so use it as you see fit, or write your own):
package hs.mediasystem.util;
import java.util.ArrayList;
import java.util.HashMap;
import java.util.List;
import java.util.Map;
import javafx.beans.InvalidationListener;
import javafx.beans.Observable;
import javafx.beans.WeakInvalidationListener;
import javafx.beans.property.Property;
import javafx.beans.value.ObservableValue;
public class WeakBinder {
private final List<Object> hardRefs = new ArrayList<>();
private final Map<ObservableValue<?>, WeakInvalidationListener> listeners = new HashMap<>();
public void unbindAll() {
    for(ObservableValue<?> observableValue : listeners.keySet()) {
      observableValue.removeListener(listeners.get(observableValue));
    hardRefs.clear();
    listeners.clear();
public <T> void bind(final Property<T> property, final ObservableValue<? extends T> dest) {
    InvalidationListener invalidationListener = new InvalidationListener() {
      @Override
      public void invalidated(Observable observable) {
        property.setValue(dest.getValue());
    WeakInvalidationListener weakInvalidationListener = new WeakInvalidationListener(invalidationListener);
    listeners.put(dest, weakInvalidationListener);
    dest.addListener(weakInvalidationListener);
    property.setValue(dest.getValue());
    hardRefs.add(dest);
    hardRefs.add(invalidationListener);
}Let me know if this solves your problem.

How to root out memory leak with Java JNI & Native BDB 11g ?

We are testing a web application using the 32-bit compiled native 11g version of BDB (with replication) under 32-bit IBM 1.5 JVM via JNI under 64-bit RedHat Linux. We are experiencing what appears to be a memory leak without a commensurate increase in Java heap size. Basically the process size continues to grow until the max 32-process size is reached (4Gb) and eventually stops running (no core). Java heap is set to 2Gb min/max. GCs are nominal, so the leak appears to be native and outside Java bytecode.
We need to determine whether there is a memory leak in BDB, or the IBM JVM or simply a mis-use of BDB in the Java code. What tools/instrumentation/db statistic should be used to help get to root cause? Do you recommend using System Tap (with some particular text command script)? What DB stats should we capture to get to the bottom of this memory leak? What troubleshooting steps can you recommend?
Thanks ahead of time.
JE.
Edited by: 787930 on Aug 12, 2010 5:42 PM

That's troublesome... DB itself doesn't have stats that track VM in any useful way. I am not familiar with SystemTap but a quick look at it seems to imply that it's better for kernel monitoring than user space. It's pretty hard to get DB to leak significant amounts of memory. The reason is that it mostly uses shared memory carved from the environment. Also if you are neglecting to close or delete some object DB generally complains about it somewhere.
I don't see how pmap would help if it's a heap leak but maybe I'm missing something.
One way to rule DB out is to replace its internal memory allocation functions with your own that are instrumented to track how much VM has been allocated (and freed). This is very easy to do using the interfaces:
db_env_set_func_malloc()
db_env_set_func_free()
These are global to your process and your functions will be used where DB would otherwise call malloc() and free(). How you get usage information out of the system is an exercise left to the reader :-) If it turns out DB is the culprit then there is more thinking to do to isolate the problem.
Other ideas that can provide information if not actual smoking guns:
-- accelerate reproduction of the problem by allocating nearly all of the VM to the JVM and the DB cache (or otherwise limit the allowable VM in your process)
-- change the VM allocated to the JVM in various ways
Regards,
George

Detecting a memory leak in a process running on Solaris 5.7

Hi,
Iam running my application on solaris 5.7 platform.
I can see clear memory leak in that process.
Is there any tool or in-built functionality on solaris 5.7 to detect/fix that leak.
Thanks in advance,
Avvaru

Thanks for the reply. I already turned it off in both Safari and Spotlight preferences. However the problem persists. The process still runs, and from time to time it decides to leak all the memory as well. I really can't understand what the reason might be.
To be clear, in Safari-->Preferences-->Search-->Smart search field-->"Include Spotlight Suggestions" is un-checked. In Spotlight preferences all three "Spotlight Suggestions", "Bookmarks & History" and "Bing Web Searches" are un-checked. Process still runs under Spotlight as can be seen here:
After the last post I let it run for a while when it happened again, and below is just before i terminated it. Look at the memory usage!

How can I avoid memory leak problem ?

I use Jdev 10.1.2 . I have a memory leak problem with ADF .
My application is very large . We have at least 30 application module , each application module contain many view object
and I have to support a lot of concurrent users .
as I know ADF stored data of view object in http session .
and http session live is quite long . when I use application for a while It raise Ouf of Memory error .
I am new for ADF.
I try to use clearCache() on view object when I don't use it any more .
and call resetState() when I don't use Application Module any more
I don't know much about behavior of clearCache() and resetState() .
I am not sure that It can avoid memory leak or not .
Do you have suggestion to avoid this problem ?

ADF does not store data in the HTTP session.
See Chapter 28 "Application Module State Management" in the ADF Developer's Guide for Forms/4GL Developers on the ADF Learning Center at http://download-uk.oracle.com/docs/html/B25947_01/toc.htm for more information.
See Chapter 29 "Understanding Application Module Pooling" to learn how you can tune the pooling parameters to control how many modules are used and how many modules "hang around" for what periods of time.

App performs inconsistantly...how to check for memory leaks?

I (was!) almost done with an app to submit to the app store but when I tested the final build on my iPad it started performing strangely.
It had never done this before, but it's crashing almost CONSTANTLY. Every few seconds or so. I haven't changed much of anything since the last time I tested it and I can't really figure out what would cause it to crash so much.
From the menu you can choose a button to go to other frames in the main timeline. Half of those work, half of them make the app crash. This is new and I think I might have a memory leak somewhere because it's consistently inconsistent.
At first I thought I had a stack overflow somewhere; however, testing in mobile debug mode doesn't cause any errors or even warning output messages - nothing. It runs just fine.
So is it possible that the problem is stemming from hardware? I can't imagine this app would be using up too much memory - it is full of vectors, but there's nothing more complicated than touch and drag because it's just a puzzle.
How would I test something like this if debug mode isn't throwing any exceptions?

You may want to try MonsterDebugger (free, open source). It also serves as a profiler. You might be able to get a better understanding on what your app is doing (and why it might be breaking on the iPad).

How to solve UIDatePicker memory leak problem in iOS 5.1?

Hi,
I am developing an application in Xcode 4.3.2. There is a page where I am using UIDatePicker to enter DOB (Date of birth). The functionality works fine but I am facing problem of memory leaks with UIDatePicker. I am followed many forums but all they say is, UIDatepicker is faulty component; it is the bug in ios 5.1. Is that correct? Should I not use this component in my application? Can anybody guide me what should I do?
Regards
Anil

It's up to date. Anything else?

How to free up memory leak?

Hi Friends,
DB = Oracle 10g
Apps=EBS R12
OS = Linux RHEL 4
My server physical RAM is 4Gb.
At fresh reboot the Memory consumption is 80M only.
When I started DB is goes up to 800Mb memory usage.
When I started Apps is goes up to 2Gb.
But when I shutdown DB and Apps I am expecting is will go back to 80M.
but still the 2Gb is used.
When I startup DB and Apps again the usage goes up to 4Gb.
How do I resolve this please :(
What is command in Linux that will list all memory usage per module/apps sorted memory size ?
Thanks a lot

have a look at this thread:
tuning Linux semaphores

How to solve this memory leak???

I've noticed as of recently that my 13" mbp, 2.7ghz, early 2011, OSX Lion 10.7.5 has been getting less and less memory as I use it throughout the day.
I upgraded my memory from 4gb to 8gbs recently thinking this would be the solution, but its still the same problem occuring.
Help please, I have no idea how to solve!

It's up to date. Anything else?

How Do I identify a Memory Leak?

I know this is not enought information but...
My application used 3 hashtables in a Data object to store application data. I retrive this data from an XML file. I also download XML and merge it into andother XML file. Some of these hashtables contain object that contain more hashtables.
I am trying to use JProbe Profiler to find the leak. I notice the memory increases when I load the data into objects and put them into hash tables. I also notice an increase when I am merging newly downloaded XML into the current XML data. I am useing Nodelists and NOdes to get the data.
Does anyone have any tips or suggestion on how to find a memory leak?
Thank you,
Al

7 MB of XML file, loaded into a DOM? You realize, of
course, that DOMs are data structures that take up
memory, too, right?Yes I do. I load it into DOM then go through the Document and put the XML data into a custom file format (seialized objects). The data is sent from the server in the form of XML.
There's your problem. If you load X MB into memory
and keep it around as long as your application is
running, that's not necessarily a "leak". Yeah I realized this but it seems that if I load 7MB of my objects into hashtables my memory usage shouldn't increase by 30MB. I just assumed it would stay around 10MB.
The real issue is trading memory for CPU. If keeping
them in memory is crashing your app with OOM errors,
by all means just keep those objects in memory for as
long as you need them and then let the GC clean them
up. Don't put them in a Map, just let them have a
narrow method scope. Recreate them every time you
need them. Yes I'm beginning to see the light.
I'm sure you're thinking, "But recreating them will
cost CPU and slow down my app!" Yes, that's why it's
a tradeoff between memory and CPU.
Do you really need the whole document in memory at
once? If you're just cherry picking a few values out
of it, maybe you can use a SAX parser instead of a
DOM.
No I don't need all that data at once.
>
Either you're screwed or you need to rethink this.
You'd better hope that it's really not needed at
every moment for every client.After rethinking, I'm going to keep the serialized data in the file until I need it. Rather than searching through a hashtable for the object I want I will search the file.
What are your JVM settings for -Xms and -Xmx? Maybe
increasing those will do the trick.I did increase them and it helped a little, but I can still get a OutOfMemoryError if I try.
Are you sure these are the culprits? Have you
profiled the app to see if there are other sources of
leaks?Yes I have been profiling the application. It seems that there is a memory leek. I am just having trouble identifing it. I know the memory increases at 2 points. When I load the XML data into the DOM and when I load the DOM into my Custom object and then into the Hashtables.
Last resort? Buy more physical memory. It's cheap.I think I am going to keep unused data out of memory and access it form a file. After I make this change I think my memory usage will drop alot. I will profile again after I have this change made to see if there is a memory leak.
Thanks for you help,
Al

Memory leak issue in tuxedo server

Hi,
I am facing memory issues in our tuxedo application.One of the server memory utilization is gradually increasing after reboot. As a temporary solution , we are making restart of that particular server every week.
I have installed Valgrind tool.i am trying to call the service ( Around 800 services are part of application ) which i suspect causing memory leak through ud script. like below.
valgrind ud32 < servicename.ud
but it is not giving memory leak details in actual service instead it is giving details about ud.
Could you guys suggest how can we find memory leak in tuxedo services.
Thanks
Ramana

HI Sir,
Thanks a lot for your response.
I ran the comand as mentioned by you in your post. but i haven't shutdown the server bcoz it is prod environment.
It logged huge output to the log file.
Below is the summary of the logfile.
==00:00:00:05.383 23666== LEAK SUMMARY:
==00:00:00:05.383 23666== definitely lost: 7,105 bytes in 7 blocks
==00:00:00:05.383 23666== indirectly lost: 3,383 bytes in 18 blocks
==00:00:00:05.383 23666== possibly lost: 28,653 bytes in 9 blocks
==00:00:00:05.383 23666== still reachable: 529,286 bytes in 288 blocks
==00:00:00:05.383 23666== suppressed: 0 bytes in 0 blocks
==00:00:00:05.383 23666==
==00:00:00:05.383 23666== For counts of detected and suppressed errors, rerun with: -v
==00:00:00:05.383 23666== Use --track-origins=yes to see where uninitialised values come from
==00:00:00:05.383 23666== ERROR SUMMARY: 107 errors from 72 contexts (suppressed: 10 from 10)
But I am not able to understand the output logfile. Could you please let me know how can we analyze the logfile.
Thanks in advance.

Finding Memory leaks in C using Visual Studio 2013

I am using Visual Studio 2013, and taking a course in C programming and we started
talking about memory bugs detaction, in particular memory leaks, and how to detect them using Valgrind on Linux.
I want to know if there is a way to detect such memory leaks using VS 2013. I tried searching online but it just leads to lots of articles and blogs and msdn posts and blogs with lots of words like dump files, and analyzing them, etc which is weird
because:
1) It sounds so complex, convoluted and unintuitive it is actually hard to comprehend it.
2) The main reason this is weird is due to the fact that VS is the most advanced IDE around and Microsoft spends so much money on in, yet from what I have read it seems that there is no simple way to use VS to detect memory leaks
- certainly no way that's as simple as Valgrind where I only have to compile the program and run the command valgrind -leaks-check=yes ProgramName
and it simply prints to me all location it thinks there is a memory leak and describes the error (like not freeing memory after allocating it with malloc hence having "dead" memory after the program finishes, or accessing memory that's out of
the array bounds)
So my question is how to use VS 2013 in order to achieve the same results and to find out First in high level if there are memory leaks in the program, and Second - to detect in a simple manner where those leaks are- preferably without involving dump files
(not that I know how to use them anyway in VS).

Hi MicrosoftLaw,
Thanks for your post.
Based on your issue, if you want to check if there are memory leaks in the C program from this VS2013. I suggest you could try to find Memory Leaks Using the CRT Library, for more information:
https://msdn.microsoft.com/en-us/library/x98tx3cf.aspx?f=255&MSPPError=-2147217396
In addition, I find a similar thread about this issue, please you refer the Dusty's suggestion to check this issue.
http://stackoverflow.com/questions/45627/how-do-you-detect-avoid-memory-leaks-in-your-unmanaged-code
I did some research about this issue, I found that the Visual Leak Detector extension tool is a free, robust, open-source memory leak detection system for Visual C++. So if possible, I suggest you could try to use the Visual Leak Detector extension
tool to find the memory leak for Visual C/ C++ program.
https://visualstudiogallery.msdn.microsoft.com/7c40a5d8-dd35-4019-a2af-cb1403f5939c
However, if you have any issue about how to use this Visual Leak Detector extension tool to find the memory leak for C program. I suggest you could directly to write a review to this REVIEW tab in Visual Leak Detector site.
Best Regards,
We are trying to better understand customer views on social support experience, so your participation in this interview project would be greatly appreciated if you have time. Thanks for helping make community forums a great place.
Click
HERE to participate the survey.

Memory leaks with Third party classes

Hello all,
This is fairly known problem. But I guess, my problem gives a new dimention to the problem.
I have an application which is developed by me. This application ideally needed to use third party classes ( obviously no source code is supplied ). These third party classes provide extra functionality required.
The problem is, when I don't use third party classes in my application, every thing is fine. When I include third party classes, I am having memory leaks.
Then I tried to investigate memory leaks with Optimizeit tool. It is new to me. As of now, I understood, we can identify where the memory leaks are occuring.
finally the problem is, in order to solve this, I need some patches in the code. But I don't have source code for those classes. How to solve this problem?
For example,
I use a third party classes in my code like this,
ThirdPartyMemoryLeakClass obj = new ThirdPartyMemoryLeakClass();
This 'obj' is made static, as it takes lot of time to create this object. Obviously this object contains several references to other objects, which I can't control.
In the process of reusing this object, I am getting memory leaks.
Any ideas regarding, how one has to deal this type of situations? What are the issues involved with this case? Are there any similar problems, which have been solved? are most welcome.
many thanks for your time.
Madhav

Decompile it using jad. Find leak.Yes, I too got the idea and tried to decompile those classes and recompile. I had some problems while recompiling. Is this is the only way to get rid of this problem?
I was refering to powersoft.datawindow.DataStore class. Does any body here has worked on these?
Can you suggest me how to find the memory leak causes? if you were needed to find out memory leak causes, what would be your approach?
Madhav

Possible memory leak in JSF � help needed

Hi,
We are developing a JSF application using the following:
SUN JSF 1.1_02-b08
Tomahawk 1.1.3
Tomcat 5.5
Our application includes pages with lists; these lists are held and extracted from backing beans.
We are using these beans in request scope with <t:saveState> tag (to avoid holding the lists in the session all the time).
We tested the application using JProbe and found out that each bean that was used in the saveState was added to the view (to the session) , each time we entered the page , and the bean was not released until the end of the session (instead of only one instance of the backing bean).
After I found this problem, I added the following context params to the web.xml:
com.sun.faces.NUMBER_OF_VIEWS_IN_SESSION = 1 (instead of the default 15)
com.sun.faces.NUMBER_OF_VIEWS_IN_LOGICAL_VIEW_IN_SESSION = 1 (instead of the default 15)
We tested the application again, and now no unnecessary instances of the beans were held in the session (in the view).
BUT , now when we use the browser �Back� button, and returning to a page with a list, when selecting an item from the list, no action is performed (the list in the server side does not exist).
My questions are:
1)What is the meaning (difference) of each of the context param : NUMBER_OF_VIEWS_IN_SESSION , NUMBER_OF_VIEWS_IN_LOGICAL_VIEW_IN_SESSION
2)What are the recommended values for these parameters, to minimize session memory, and to enable the usage of the browser �Back� button.
Thank you all,
Lior.

Me too.
Please somebody knows how can we avoid memory leak problems in JSF 1.1 ?
Thansk in advanced

How to detect a memory leak?

Similar Messages

Maybe you are looking for