Killing 1.4.1 parent jvm doesn't kill children also

[Joel Lindsey]

We have noticed a small behavior change in 1.4.1 vs. 1.3.1. We have a
java manager process that launches and watches child java processes.
It uses System.Runtime.exec() to launch the child processes. With the
1.3.1 JVM, if you kill the manager process (with task manager or a
ctrl-C) it takes the child processes down with it (this is what we
want). With the 1.4.1 JVM, if you kill the manager the child
processes hang around.

Does anyone know of a way to get the 1.4.1 Windows JVM to behave like
the 1.3.1 jvm?

Thanks

 

[Steve W. Jackson]

:We have noticed a small behavior change in 1.4.1 vs. 1.3.1. We have a
:java manager process that launches and watches child java processes.
:It uses System.Runtime.exec() to launch the child processes. With the
:1.3.1 JVM, if you kill the manager process (with task manager or a
:ctrl-C) it takes the child processes down with it (this is what we
:want). With the 1.4.1 JVM, if you kill the manager the child
rocesses hang around.
:
oes anyone know of a way to get the 1.4.1 Windows JVM to behave like
:the 1.3.1 jvm?
:
:Thanks

If the 1.3.1 JVM behaved this way, then it's an error. Using the exec()
method spawns a completely separate process, not one that's within the
same JVM (assuming the command executed is indeed Java) as that exec()
statement. So unless the OS and/or processes involved need to kill
spawn child processes when the parent dies, it's got no bearing on the
JVM at all.

= Steve =