-
Bug
-
Resolution: Duplicate
-
P3
-
None
-
8u45
-
x86_64
-
windows_7
FULL PRODUCT VERSION :
java version "1.8.0_45"
Java(TM) SE Runtime Environment (build 1.8.0_45-b14)
Java HotSpot(TM) 64-Bit Server VM (build 25.45-b02, mixed mode)
also:
java version "1.7.0_65"
Java(TM) SE Runtime Environment (build 1.7.0_65-b19)
Java HotSpot(TM) 64-Bit Server VM (build 24.65-b04, mixed mode)
ADDITIONAL OS VERSION INFORMATION :
Microsoft Windows [Version 6.1.7601]
A DESCRIPTION OF THE PROBLEM :
Our customer reported a dangerous regression when using Ant to start background processes from a Windows command prompt, in which a child process that was started by a (now-completed) invocation of ant/Java get unexpectedly terminated when the command prompt is closed or receives a Ctrl+C.
I can repro with a trivial invocation of Runtime.exec, which is showing a clear regression in behaviour between Java 1.7.0_45 to _65 (and is also broken in the same way on the latest 1.8.0_45). I also tried using ProcessBuilder with various inheritance options and wasn't able to work around it, so it seems there is now no way to safely exec a background child process in Java from a windows command prompt.
Very likely this was broken by the fix for either http://bugs.java.com/view_bug.do?bug_id=7147084 or http://bugs.java.com/view_bug.do?bug_id=8007454 (which both went into update 60).
It's not obvious to me from the MSDN doc on CreateProcess https://msdn.microsoft.com/en-us/library/windows/desktop/ms682425%28v=vs.85%29.aspx?f=255&MSPPError=-2147217396 what the solution is, but the handle inheritance changes are by far the most likely culprit. I don't know whether setting the DETACHED_PROCESS process creation flag would help.
REGRESSION. Last worked in version 7u75
STEPS TO FOLLOW TO REPRODUCE THE PROBLEM :
- Open a Windows command prompt
- use java.exe to execute a trivial java program that uses Runtime.exec to start a long-lived background process - doesn't make a difference whether it generates output (e.g. ping) or not (e.g. sleep); see sample program below
- after the Java process has terminated, the newly created child process (e.g. ping) naturally continues running as expected
- now press Ctrl+C (at the C:\> prompt), or close the parent command prompt window
EXPECTED VERSUS ACTUAL BEHAVIOR :
EXPECTED -
The background process should continue running unaffected (can use task manager to verify this)
ACTUAL -
The background process terminates.
This is a regression in behaviour - the child process continues to run fine with 1.7.0u45, but does not with u65
REPRODUCIBILITY :
This bug can be reproduced always.
---------- BEGIN SOURCE ----------
public class Exec
{
public static void main(String[] args) throws Exception
{
Runtime.getRuntime().exec(new String[]{"ping", "-n", "10000", "localhost"});
}
}
---------- END SOURCE ----------
java version "1.8.0_45"
Java(TM) SE Runtime Environment (build 1.8.0_45-b14)
Java HotSpot(TM) 64-Bit Server VM (build 25.45-b02, mixed mode)
also:
java version "1.7.0_65"
Java(TM) SE Runtime Environment (build 1.7.0_65-b19)
Java HotSpot(TM) 64-Bit Server VM (build 24.65-b04, mixed mode)
ADDITIONAL OS VERSION INFORMATION :
Microsoft Windows [Version 6.1.7601]
A DESCRIPTION OF THE PROBLEM :
Our customer reported a dangerous regression when using Ant to start background processes from a Windows command prompt, in which a child process that was started by a (now-completed) invocation of ant/Java get unexpectedly terminated when the command prompt is closed or receives a Ctrl+C.
I can repro with a trivial invocation of Runtime.exec, which is showing a clear regression in behaviour between Java 1.7.0_45 to _65 (and is also broken in the same way on the latest 1.8.0_45). I also tried using ProcessBuilder with various inheritance options and wasn't able to work around it, so it seems there is now no way to safely exec a background child process in Java from a windows command prompt.
Very likely this was broken by the fix for either http://bugs.java.com/view_bug.do?bug_id=7147084 or http://bugs.java.com/view_bug.do?bug_id=8007454 (which both went into update 60).
It's not obvious to me from the MSDN doc on CreateProcess https://msdn.microsoft.com/en-us/library/windows/desktop/ms682425%28v=vs.85%29.aspx?f=255&MSPPError=-2147217396 what the solution is, but the handle inheritance changes are by far the most likely culprit. I don't know whether setting the DETACHED_PROCESS process creation flag would help.
REGRESSION. Last worked in version 7u75
STEPS TO FOLLOW TO REPRODUCE THE PROBLEM :
- Open a Windows command prompt
- use java.exe to execute a trivial java program that uses Runtime.exec to start a long-lived background process - doesn't make a difference whether it generates output (e.g. ping) or not (e.g. sleep); see sample program below
- after the Java process has terminated, the newly created child process (e.g. ping) naturally continues running as expected
- now press Ctrl+C (at the C:\> prompt), or close the parent command prompt window
EXPECTED VERSUS ACTUAL BEHAVIOR :
EXPECTED -
The background process should continue running unaffected (can use task manager to verify this)
ACTUAL -
The background process terminates.
This is a regression in behaviour - the child process continues to run fine with 1.7.0u45, but does not with u65
REPRODUCIBILITY :
This bug can be reproduced always.
---------- BEGIN SOURCE ----------
public class Exec
{
public static void main(String[] args) throws Exception
{
Runtime.getRuntime().exec(new String[]{"ping", "-n", "10000", "localhost"});
}
}
---------- END SOURCE ----------
- duplicates
-
JDK-8067748 (process) Child is terminated when parent's console is closed [win]
- Closed