glassfish crash caused by libdbus on debian squeeze - glassfish

I'm running glassfish v3.1.1 build 12 using java build 1.6.0_29-b11 on linux debian squeez (6.0.4).
I'm running a webapp which is calling an external soap webservice.
Every now and then the jvm crashes. I can reproduce this crash by executing 20 threads simultaneously.
The hs_err_pid crash dump file shows that libdbus is the problematic library. Here's the interesting part of the crash dump
Stack: [0x00007f3fef0ff000,0x00007f3fef200000], sp=0x00007f3fef1f7058, free space=992k
Native frames: (J=compiled Java code, j=interpreted, Vv=VM code, C=native code)
C [libdbus-1.so.3+0x28de0]
Java frames: (J=compiled Java code, j=interpreted, Vv=VM code)
j sun.net.spi.DefaultProxySelector.getSystemProxy(Ljava/lang/String;Ljava/lang/String;)Ljava/net/Proxy;+0
j sun.net.spi.DefaultProxySelector.access$100(Lsun/net/spi/DefaultProxySelector;Ljava/lang/String;Ljava/lang/String;)Ljava/net/Proxy;+3
j sun.net.spi.DefaultProxySelector$2.run()Ljava/lang/Object;+151
v ~StubRoutines::call_stub
J java.security.AccessController.doPrivileged(Ljava/security/PrivilegedAction;)Ljava/lang/Object;
j sun.net.spi.DefaultProxySelector.select(Ljava/net/URI;)Ljava/util/List;+273
j sun.net.www.protocol.http.HttpURLConnection.plainConnect()V+187
j sun.net.www.protocol.https.AbstractDelegateHttpsURLConnection.connect()V+9
j sun.net.www.protocol.http.HttpURLConnection.getOutputStream()Ljava/io/OutputStream;+134
j sun.net.www.protocol.https.HttpsURLConnectionImpl.getOutputStream()Ljava/io/OutputStream;+4
j com.sun.xml.ws.mex.client.HttpPoster.post(Ljava/lang/String;Ljava/lang/String;Ljava/lang/String;)Ljava/io/InputStream;+60
as you can see the actual HttpPoster.post is being called after which the crash occurs. The sun.net.spi.DefaultProxySelector is part of rt.jar so this is part of the default jvm installation.
I tried jdk7 but also here the crash occurs.
Then I installed openjdk which also crashed.
Then I installed dbus-java which is the java implementation of dbus (using apt-get). Normally this library is only used when certain jar files are put explicitly in the path. But after installing this library the crashes don't occur anymore. So it looks like openjdk is using this library (assumption)
I talked to the maintainer of dbus-java and also he is not aware that sun jdk or openjdk is using dbus. So we are both quite confused.
The crashes don't occur anymore but now the admin interface of glassfish stops working. Googling shows that this is a known issue and the solution is to start using sun jdk. Aaargh.
Have you seen this problem before? Is there a solution?
UPDATE:
the DefaultProxySelector.getSystemProperty method is the last method that is called before a native method is called. Looking at the code of DefaultProxySelector I found that this method should only be called if the java.net.useSystemProxies=true. But this property is not set (no -D setting in the jvm settings) nor is it true in the $JAVA_HOME/jre/lib/net.properties file. But after debugging it shows this boolean IS true.
So the big question now is : why is this boolean true?
UPDATE 2:
The problem was caused by the updatetool configuration. See bug report GLASSFISH-18360 for details.

Related

Moonscript CLI launch error with Lua 5.3 on OSX

I'm on OS X 10.11.1 running the latest version of X Code. I installed Lua 5.3 today from source with no problem. Seems to work correctly. No errors installing luarocks and then using it to install moonscript. However, both moon and moonc give the following error when invoked from the command line:
$ moon
/usr/local/bin/lua: /usr/local/share/lua/5.3/alt_getopt.lua:24: attempt to call a nil value (global 'module')
stack traceback:
/usr/local/share/lua/5.3/alt_getopt.lua:24: in main chunk
[C]: in function 'require'
/usr/local/lib/luarocks/rocks/moonscript/0.4.0-1/bin/moon:2: in main chunk
[C]: in ?
Looking at alt_getopt.lua shows that it is indeed using the now-defunct 'module' keyword. The GitHub project for this module appears to be dead with no action in the past 3 years.
How should I proceed to get Moonscript working on my system?
Yes, as you noted, module has been deprecated for quite a while and was only working with earlier versions because of compatibility switches.
You can comment out the line 24 in alt-getopt and add the following line to the end of the alt-getopt.lua file: return {get_opts = get_opts, get_ordered_opts = get_ordered_opts}.
This should make this module Lua 5.2/5.3 compatible (I haven't checked if there are any other compatibility issues with using Lua 5.3 though).

java.lang.NoClassDefFoundError: Could not initialize class org.apache.activemq.util.IdGenerator

I am trying to use ActiveMQ 5.10.0 with SoapUI 4.6 and Hermes 1.14. I get the error below when I try and add a queue. I presume Hermes can't find the type IdGenerator in any of the loaded jars. Which are:
activemq-client-5.10.0.jar
geronimo-j2ee-management_1.1_spec-1.0.1.jar
geronimo-jms_1.1_spec-1.1.1.jar
Does anyone know where this class is defined? I looked for activemq-util.jar in the binary distribution but I did not find such a file.
Error:
java.lang.NoClassDefFoundError: Could not initialize class org.apache.activemq.util.IdGenerator
at org.apache.activemq.ActiveMQConnectionFactory.getClientIdGenerator(ActiveMQConnectionFactory.java:969)
at org.apache.activemq.ActiveMQConnectionFactory.createActiveMQConnection(ActiveMQConnectionFactory.java:363)
at org.apache.activemq.ActiveMQConnectionFactory.createActiveMQConnection(ActiveMQConnectionFactory.java:331)
at org.apache.activemq.ActiveMQConnectionFactory.createActiveMQConnection(ActiveMQConnectionFactory.java:303)
at org.apache.activemq.ActiveMQConnectionFactory.createQueueConnection(ActiveMQConnectionFactory.java:260)
at hermes.impl.jms.ConnectionManagerSupport.createConnection(ConnectionManagerSupport.java:147)
at hermes.impl.jms.ConnectionManagerSupport.createConnection(ConnectionManagerSupport.java:92)
at hermes.impl.jms.ConnectionSharedManager.reconnect(ConnectionSharedManager.java:81)
at hermes.impl.jms.ConnectionSharedManager.connect(ConnectionSharedManager.java:91)
at hermes.impl.jms.ConnectionSharedManager.getConnection(ConnectionSharedManager.java:104)
at hermes.impl.jms.ConnectionSharedManager.getObject(ConnectionSharedManager.java:142)
at hermes.impl.jms.ThreadLocalSessionManager.connect(ThreadLocalSessionManager.java:190)
at hermes.impl.jms.ThreadLocalSessionManager.getSession(ThreadLocalSessionManager.java:570)
at hermes.impl.jms.AbstractSessionManager.getDestination(AbstractSessionManager.java:460)
at hermes.impl.DefaultHermesImpl.getDestination(DefaultHermesImpl.java:367)
at hermes.browser.tasks.BrowseDestinationTask.invoke(BrowseDestinationTask.java:141)
at hermes.browser.tasks.TaskSupport.run(TaskSupport.java:175)
at hermes.browser.tasks.ThreadPool.run(ThreadPool.java:170)
at java.lang.Thread.run(Unknown Source)
This is very old but in case anyone else is trying to get this spun up... don't use the built in HermesJMS that comes with SoapUI. Apparently there's a bug in it that doesn't play nice with ActiveMQ v5.8 and following [I tried 5.11 & 5.13 and had the issue. The cheating fix is to install the standalone [I had to get it from sourceforge].
The sourceforge jar is installed with [assuming version 1.14]: java -jar hermes-installer-1.14.jar
Once installed you can tie this version to soapui or launch it with the bat/sh file. I still had issues with ActiveMQ version 5.13 but version 5.11 worked for me.
The IdGenerator class is located in the activemq-client jar. Here is the result of a search in the source tree:
/activemq-client/src/main/java/org/apache/activemq/util/IdGenerator.java
There's been no recent changes so I'd guess that the error is misleading and that it's actually some other missing dependency that's being loaded when the class is created. Is there more information in the log or a 'caused by exception' ?
For whom it may interest. This is kind of common issue I come accross from time to time. I call it jar version incompatibility. I was getting exception in title and other funny exceptions when using latest (but not actively developed) hermes 1.14 and currently latest Apache Active MQ 5.14. I've found out after a long struggle that it is fixed by using an older version of Active MQ - like 5.3.

Java 7 Migration - New JBoss Serialization-related exception

I'm trying to upgrade an application to Java 7 from Java 6 and I'm getting a new exception that isn't present when running the application under Java 6.
Here's the first section of the stack trace:
java.lang.ClassCastException: java.lang.String cannot be cast to org.jboss.serial.finalcontainers.IntegerContainer
at org.jboss.serial.objectmetamodel.DataContainer$DataContainerInput.readInt(DataContainer.java:1044)
at org.jboss.serial.persister.RegularObjectPersister.readSlotWithFields(RegularObjectPersister.java:310)
at org.jboss.serial.persister.RegularObjectPersister.defaultRead(RegularObjectPersister.java:273)
at org.jboss.serial.persister.RegularObjectPersister.readData(RegularObjectPersister.java:241)
at org.jboss.serial.objectmetamodel.ObjectDescriptorFactory.readObjectDescriptionFromStreaming(ObjectDescriptorFactory.java:412)
at org.jboss.serial.objectmetamodel.ObjectDescriptorFactory.objectFromDescription(ObjectDescriptorFactory.java:82)
at org.jboss.serial.objectmetamodel.DataContainer$DataContainerInput.readObject(DataContainer.java:845)
at org.jboss.serial.persister.RegularObjectPersister.readSlotWithFields(RegularObjectPersister.java:353)
at org.jboss.serial.persister.RegulrObjectPersister.defaultRead(RegularObjectPersister.java:273)
at org.jboss.serial.persister.RegularObjectPersister.readData(RegularObjectPersister.java:241)
at org.jboss.serial.objectmetamodel.ObjectDescriptorFactory.readObjectDescriptionFromStreaming(ObjectDescriptorFactory.java:412)
at org.jboss.serial.objectmetamodel.ObjectDescriptorFactory.objectFromDescription(ObjectDescriptorFactory.java:82)
at org.jboss.serial.objectmetamodel.DataContainer$DataContainerInput.readObject(DataContainer.java:845)
at org.jboss.serial.persister.ObjectInputStreamProxy.readObjectOverride(ObjectInputStreamProxy.java:68)
at java.io.ObjectInputStream.readObject(ObjectInputStream.java:364)
at java.util.ArrayList.readObject(ArrayList.java:733)
at sun.reflect.GeneratedMethodAccessor352.invoke(Unknown Source)
I found this JBoss bug, which is supposedly fixed...
https://issues.jboss.org/browse/JBSER-128
but the version it's fixed in (1.0.6.FINAL) hasn't been released yet:
https://issues.jboss.org/browse/JBSER
Has anybody run into this that could suggest a workaround?
The application runs in JBoss 6.0.0.
PS I am aware that JBoss 6 is past its EOL and not tested with Java 7.
Well, you could build the release yourself from the tag in the SVN repo. Or if you're really trusting (or want to do a quick experiment) download the binary I built (here, won't promise they stay online though).
Just replace all occurrences of the jar in the JBoss distribution and don't forget to enable the fix with system property
-Dorg.jboss.serial.sync_binary_formats=true
As highlighted here
https://issues.jboss.org/browse/JBSER-128?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
Anyhow for me this worked for Vectors at least (there's other potential problems, but not being able to transfer CCHashMaps I'll survive ;) ).
I found a workaround solution for my problem. Since Java 7 changed the serialization behavior of Vector, I changed to ArrayList (which did not change) and the problem has been resolved.
Hope this helps somebody out there struggling with the same issue.

JNI UnsatisfiedLinkError- how do I load libraries correctly?

I'm trying to work with the Java sample Database program from the CardScan SDK.
I am working with files located in Java/JNI and Java/Database. The program must be run with a 32 bit JRE. I was able to do so on a 64 bit machine by uninstalling Java and installing the 32 bit version, then re-adding the system path for Java. I can run the program and interface with a CardScan database file (.cdb) successfully by double clicking the SDKData.bat file, but when I open the source files for editing and edit the Java.library.path to include the required library (CRTK_JNI.dll), I get UnsatisfiedLinkErrors everywhere:
Exception in thread "main" java.lang.UnsatisfiedLinkError: sdkdata.CRTK.CRTK_Init([I)I
at sdkdata.CRTK.CRTK_Init(Native Method)
at sdkdata.CRTK.(CRTK.java:239)
at sdkdata.SDKData.(SDKData.java:97)
at sdkdata.SDKData.main(SDKData.java:643)
Java Result: 1
Presumably this is happening because the library is not loading properly.
What do I need to do to run and edit the program at full capacity (with all the native functions from CRTK_JNI in working order)?
Presumably this is happening because the library is not loading properly.
On the contrary. The library load is complete. You aren't getting that from a System.load()/loadLibrary() call, you are getting the error when calling your native method, the one that should have the signature:
package sdkdata;
public class CRTK
{
public native int CRTK_Init(int[]);
}
So it isn't there, or you have changed the signature without regenerating the .h and .c files, or you have manually mangled the declaration some other way.
Post your code.
To clarify, this Java sample program is officially unsupported by the CardScan API - it was a bad idea to try to use the API with an unsupported language relying solely on an experimental implementation. I ended up using one of the supported languages (Visual Basic) to work with the SDK; if anyone looking at this question happens to be struggling with using the CardScan API, here is my VB implementation on Github.

Objective C error messages

I'm trying to debug an objective C program. It used to run, and I'm not sure what changes could have broken it, but it no longer runs at all. In the debugger, I now get:
.
.
.
unable to read unknown load command 0x80000022
unable to read unknown load command 0x80000022
unable to read unknown load command 0x80000022
Program loaded.
sharedlibrary apply-load-rules all
run
Error calling thread_get_state for GP registers for thread 0x16131lx506^error,msg="error on line 207 of \"/SourceCache/gdb/gdb-908/src/gdb/macosx/i386-macosx-nat-exec.c\" in function \"fetch_inferior_registers\": (os/kern) invalid argument (0x4)\n"
I've seen several places that claim te explain the 1st warning, but none of them seem appropriate. Nothing online claims to explain the error.
Any clues as to what these errors mean in conjunction?
ETA: I just upgraded to 10.6, so it may have started then, although I'm compiling to target 10.4...
Is this relevant? From the Xcode release notes:
Note: GCC 4.2 cannot be used with the
Mac OS X 10.4u SDK. If you want to
build targets using the 10.4u SDK on
Xcode 3.2, you must set the Compiler
Version to GCC 4.0