kriegaex edited a comment on pull request #355: URL: https://github.com/apache/maven-surefire/pull/355#issuecomment-873703758
> You did not take my commit with the fix. Your HEAD commit in your repository contains your tests without the fix. So it is quite irrelevant. What are you talking about?  You should read my PR correctly. The PR of course builds on top of your fixes, I branched it off of master. Please revert and merge my PR as prior art instead. I did this work, you were simply recreating it. These are my commits and I want them merged. If something is wrong with my PR, we can fix them, of course. The other branch without the fixes simply shows what would happen if * we overlooked an edge case or * accidentally the problem gets re-introduced in the future by someone else's change or refactoring. This IT (like all tests) is meant to do two things: * verify some assumptions about how the application should behave * reproduce a problem, spotting regressions, in case it does not behave as expected So, testing what would happen if the test actually _would_ run into a timeout due to a regression is very much relevant. I simply created the other branch for your convenience, so you can reproduce the problem with hanging JVMs not being cleaned up properly. That other branch is not a PR branch, this one is! > Nobody is watching Windows task manager in order to send some proves because this is false representation. Of course not, normally people just run the IT, relying on it behaving correctly. Cleaning up resources and fixtures is part of a test's correct behaviour, and I manually verified that for you. I noticed that resource clean-up is broken, which is why it should be fixed, because otherwise people **not** looking at a task manager or `htop` would never notice what is going on, until all the running processes from repeatedly running tests would cause out of memory errors on their workstations or CI servers. > Instead, the developers use the command-line So the fact that I used a GUI tool from SysInternals disqualifies me as a developer? It proves what? That you are a real developer and I am not? Can you be any more condescending? Different developers use different tools. If the tool reveals the correct information, why does it matter if I started it from a console or from an IDE? Besides, **I did run the test from Maven(!),** simply used a GUI tool to watch what Maven does live. In that tool, I cannot just see the end result but also what happens during the test: First, each of the two Java process groups are indirect children of sh.exe (my Git Bash), which starts the Maven build. Then, each time one of the two tests is interrupted due to the timeouts, the subprocesses lose their parent (of course) and now move up the process hierarchy, hanging there as zombies without parents. Your `wmic` command only shows a subset of the same information, because the fact is missing that it is not just Java processes but a group of 4 processes per test case. My simple screenshot illustrates that way better. You made a similar argument about my Process Explorer screenshot in the original Jira ticket, trying to fend off my information as irrelevant, claiming the Java agent to be the problem instead of Surefire. Like back then, you are also wrong in this case: The problem is real, my screenshot illustrates it you can easily reproduce it in my test branch. Of course, currently there is no known regression, but no software is perfect and regressions can always occur. To ensure that the ITs clean up their resources in case they fail due to a hanging JVM (which is exactly what they test should **not** happen), is part of the job of writing tests. So the job is not done yet. But if it makes you happy, I replayed again on the console for you (ran inside the IT module after a previous full build with install): ```none $ mvn -P run-its verify (...) [INFO] ------------------------------------------------------- [INFO] T E S T S [INFO] ------------------------------------------------------- [INFO] Running org.apache.maven.surefire.its.jiras.Surefire1881IT [ERROR] Tests run: 2, Failures: 0, Errors: 2, Skipped: 0, Time elapsed: 60.054 s <<< FAILURE! - in org.apache.maven.surefire.its.jiras.Surefire1881IT [ERROR] org.apache.maven.surefire.its.jiras.Surefire1881IT.testJVMLogging Time elapsed: 30.011 s <<< ERROR! org.junit.runners.model.TestTimedOutException: test timed out after 30000 milliseconds at app//org.apache.maven.surefire.its.jiras.Surefire1881IT.testJVMLogging(Surefire1881IT.java:37) [ERROR] org.apache.maven.surefire.its.jiras.Surefire1881IT.testJavaAgent Time elapsed: 30.011 s <<< ERROR! org.junit.runners.model.TestTimedOutException: test timed out after 30000 milliseconds at app//org.apache.maven.surefire.its.jiras.Surefire1881IT.testJavaAgent(Surefire1881IT.java:47) [INFO] [INFO] Results: [INFO] [ERROR] Errors: [ERROR] Surefire1881IT.testJVMLogging:37 » TestTimedOut test timed out after 30000 mil... [ERROR] Surefire1881IT.testJavaAgent:47 » TestTimedOut test timed out after 30000 mill... [INFO] [ERROR] Tests run: 2, Failures: 0, Errors: 2, Skipped: 0 (...) $ wmic PROCESS where "name like '%java%'" get Commandline CommandLine "C:\Program Files\Java\jdk-16\bin\java.exe" -Dfile.encoding=UTF-8 -classpath "C:\Program Files\Java\apache-maven-3.6.3\bin\..\boot\plexus-classworlds-2.6.0.jar" "-Dclassworlds.conf=C:\Program Files\Java\apache-maven-3.6.3\bin\..\bin\m2.conf" "-Dmaven.home=C:\Program Files\Java\apache-maven-3.6.3\bin\.." "-Dlibrary.jansi.path=C:\Program Files\Java\apache-maven-3.6.3\bin\..\lib\jansi-native" "-Dmaven.multiModuleProjectDirectory=C:\Users\alexa\Documents\java-src\maven-surefire\surefire-its\target\Surefire1881IT_testJVMLogging" org.codehaus.plexus.classworlds.launcher.Launcher -e --batch-mode -Dmaven.repo.local=C:\Users\alexa\.m2\repository org.apache.maven.plugins:maven-clean-plugin:clean -Dsurefire.version=3.0.0-M6-SNAPSHOT -Djacoco.agent=-javaagent:C:\\Users\\alexa\\.m2\\repository\\org\\jacoco\\org.jacoco.agent\\0.8.7\\org.jacoco.agent-0.8.7-runtime.jar=destfile=C:\\Users\\alexa\\Documents\\java-src\\maven-surefire\\surefire-its\\target\\jacoco-it.exec,includ es=**/failsafe/*:**/failsafe/**/*:**/surefire/*:**/surefire/**/*,excludes=**/HelpMojo.class:**/shadefire/**/*:org/jacoco/**/*:com/vladium/emma/rt/* -Dhttps.protocols=TLSv1,TLSv1.1,TLSv1.2 -nsu test "C:\Program Files\Java\jdk-16\bin\java" -verbose:class -jar C:\Users\alexa\AppData\Local\Temp\surefire2405688146490858653\surefirebooter4985087880621361102.jar C:\Users\alexa\AppData\Local\Temp\surefire2405688146490858653 2021-07-05T07-34-52_138-jvmRun1 surefire6407547456865244099tmp surefire_09695795624160661675tmp "C:\Program Files\Java\jdk-16\bin\java.exe" -Dfile.encoding=UTF-8 -classpath "C:\Program Files\Java\apache-maven-3.6.3\bin\..\boot\plexus-classworlds-2.6.0.jar" "-Dclassworlds.conf=C:\Program Files\Java\apache-maven-3.6.3\bin\..\bin\m2.conf" "-Dmaven.home=C:\Program Files\Java\apache-maven-3.6.3\bin\.." "-Dlibrary.jansi.path=C:\Program Files\Java\apache-maven-3.6.3\bin\..\lib\jansi-native" "-Dmaven.multiModuleProjectDirectory=C:\Users\alexa\Documents\java-src\maven-surefire\surefire-its\target\Surefire1881IT_testJavaAgent" org.codehaus.plexus.classworlds.launcher.Launcher -e --batch-mode -Dmaven.repo.local=C:\Users\alexa\.m2\repository org.apache.maven.plugins:maven-clean-plugin:clean -Dsurefire.version=3.0.0-M6-SNAPSHOT -Djacoco.agent=-javaagent:C:\\Users\\alexa\\.m2\\repository\\org\\jacoco\\org.jacoco.agent\\0.8.7\\org.jacoco.agent-0.8.7-runtime.jar=destfile=C:\\Users\\alexa\\Documents\\java-src\\maven-surefire\\surefire-its\\target\\jacoco-it.exec,include s=**/failsafe/*:**/failsafe/**/*:**/surefire/*:**/surefire/**/*,excludes=**/HelpMojo.class:**/shadefire/**/*:org/jacoco/**/*:com/vladium/emma/rt/* -Dhttps.protocols=TLSv1,TLSv1.1,TLSv1.2 -nsu verify "C:\Program Files\Java\jdk-16\bin\java" -javaagent:C:\Users\alexa\Documents\java-src\maven-surefire\surefire-its\target\Surefire1881IT_testJavaAgent\target/surefire-1881-java-agent-1.0-SNAPSHOT.jar -jar C:\Users\alexa\AppData\Local\Temp\surefire1253544507142731543\surefirebooter12628764455616323129.jar C:\Users\alexa\AppData\Local\Temp\surefire1253544507142731543 2021-07-05T07-35-22_653-jvmRun1 surefire11527327993772670760tmp surefire_018188072805511397606tmp ``` The corresponding screenshot from SysInternals `procexp64` when running Maven from Git Bash:  **Update:** When running from cmd.exe, there are only two processes left per test timed out method, not 4:  The difference between both shells seems to originate from the fact that Git Bash forks a cmd.exe when running Maven, so there is more stuff to clean up. Why Maven in one case (cmd.exe) can terminate the parent Java process and just not the child, but in the other case (Git Bash) cannot terminate either of them, is to be investigated. -- This is an automated message from the Apache Git Service. To respond to the message, please log on to GitHub and use the URL above to go to the specific comment. To unsubscribe, e-mail: issues-unsubscr...@maven.apache.org For queries about this service, please contact Infrastructure at: us...@infra.apache.org
