wrapper-windows-x86-32-3.5.19

Java Service Wrapper Revision History. -------------------------------------- 3.5.19 * Fix a problem in the batch file, where a space in the path to the Wrapper.exe file would cause the script to locate the Wrapper.exe file. Introduced in 3.5.18. * When running as Windows service with hiding the console of the Wrapper will cause the Wrapper to disable unnecessary logging to the console in order to enhance performance. 3.5.18 * Fix a problem, where an unclosed percentage character '%' was opening the chance of a dangling pointer in the additional java parameters. The '%' character is a special character, specifying an environment variable. * Added variable _WRAPPER_DIR the batch files to make it possible to specify any other directory where the Wrapper binary file is located. Until now the batch file and exe file had to be in the same location. Thanks and credits go to Geoff. * Added property wrapper.port.address, which makes it possible to specify a different address to bind the socket to when using the socket backend connection between the Wrapper and the JVM. Until now, the socket was always bound to the localhost loopback interface. * The script will from now on also use the update-rc.d tool for installing an application as daemon on Debian. Thanks and credits go to Haifeng. * Whenever the Wrapper is causing the JVM to be forcibly terminated, the Wrapper will make sure the JVM has actually been terminated. If it wasn't after wrapper.jvm_terminate.timeout seconds, a pending restart will be canceled and the Wrapper exit. * Reworked the way the Wrapper is processing output from the JVM, in order to increase performance on huge output on a single line and also reduce memory usage. 3.5.17 * Add a new wrapper.java.additional.default.stripquotes property to make it possible to specify the default value of wrapper.java.additional.<n>.stripquotes * Fix a bug where the timer failed to calculate the fire time when that time was more than one week in the future. This was possible for weekly timers which spanned a daylight savings time change which rolled the time back by an hour in the fall. * Fix problem in the shell script, where it might fail to remove an installed daemon after the location of the script has been changed. * Add additional advice messages when a Windows service fails to be started due to file access problems involving the Wrapper binary, configuration, or log files. * Fix a problem where the dynamic library on MacOSX was not able to load it's functions. * Added wrapper.app.parameter_file property, which works similar to the wrapper.java.additional_file property * Reduce CPU-consuption of WrapperProcess.waitFor() function 3.5.16 * (Standard, Professional) Retry failed share mappings if the target host or network is unreachable as that may be a temporary problem. * (Professional) There was a problem where the IO-redirection of a child process which got created with the WrapperManager.exec API and used the feature to run the child process in the logged on users desktop was only allowing to create a process once per second. * Include information about the base configuration file in the debug output when debugging of cacading configuration files has been enabled. * Add a check in the UNIX script to output a more descriptive error message, when the user specified in the RUN_AS_USER variable doesn't exist. * (Standard, Professional) Fix a problem where console log output was not being displayed correctly when running with the WrapperW.exe binary with the wrapper.ntservice.console property was set to true. * (Standard, Professional) Implement the wrapper.ntservice.generate_console property when using the WrapperW.exe binary so it is now possible to disable the creation of the hidden console. * Modify the way the the wrapper.ntservice.generate_console property works so it is now easier to disable the generation of the console using just this property. * Improve the message logged when the the Wrapper attempts to perform a thread dump without a valid console being available. * Add new wrapper.ping.alert.threshold and wrapper.ping.alert.loglevel properties which make it much easier to debug ping timeout issues by asking the Wrapper to log messages about ping responses which were shorter than the registered wrapper.ping.timeout, but longer than the threshold. * Add a new WrapperManager.appearSlow method which makes it easier to test how the Wrapper behaves when the JVM is being slow to respond to commands. * Add a new wrapper.disable_tests property which can be used to disable all of the testing methods of the WrapperManager class. It has always been possible to control their access with a SecurityManager, but this is simpler for most applications. * Update the default wrapper configuration file template so a restart due to a matched OutOfMemoryError filter will no longer be triggered by default if the user enables -verbose:class output. * Fix a problem on UNIX platforms where the Wrapper would fail to start if it was located on the system PATH. This had been a problem since version 3.3.0 but had gone unnoticed as the Wrapper is not usually referenced in this way. * Rework the internal flags governing the generation and hiding of the backend console on Windows so we are able to almost always obtain the console's window handle. * Cleanup some startup code to reduce duplication and make sure that more debug and warning messages are logged after the "Wrapper Started" message. * Add new wrapper.java.additional_file and wrapper.java.additional_file.stripquotes properties to make it possible to specify JVM parameters in a file. * Add support for Linux on ARM systems. * Re-Enabled the forced reloading of the SYSTEM (and if set to a specific account, the user) registry before launching the Wrapper as a service on Windows XP and 2003. This has been originally disabeld for Windows XP and 2003 since version 3.5.5. * (Standard, Professional) Fix a problem where the instance class names logged when a deadlock involving ReentrantLock instances were corrupted. The actual deadlock detection was working correctly, but this could have lead to other problems caused by the corruption. A workaround was to set the wrapper.check.deadlock.output property to SIMPLE. * (Standard, Professional) Make it possible to completely disable the details of a deadlock by setting the wrapper.check.deadlock.output property to NONE. * (Standard, Professional) Object Ids in thread dump reports were not correctly being logged as 64-bit ids on 64-bit JVMs in some cases. * Fix a problem where the source code values returned by the WrapperServiceActionEvent.getSourceCode() method were incorrect. The constant values were incorrect and have been corrected from this release. * Add new WrapperServiceActionEvent.getSourceCodeName() and WrapperServiceActionEvent.getSourceCodeName(int actionSourceCode) methods which returns a localized name of the source where the event originated. * Fix a minor problem where a couple uncommon backend packet codes were not being correctly identified by name in the debug log output. Functionally they were all working correctly. * Added property wrapper.ping.timeout.action, which will let you specify an action in case the timeout triggers. So far the only action was to restart the JVM. * Fix a problem where a JVM process was not stopped completely on a UNIX platform and stayed defunct after a forced kill until the Wrapper process itself stopped. This was especially noticeable if the JVM is frozen and the JVM is being killed forcibly. 3.5.15 * Add a new _WRAPPER_CONF_OVERRIDE setting to the Wrapper dedicated command batch files on Windows so it is now possible to