[ 
https://issues.apache.org/jira/browse/GEODE-5787?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16637641#comment-16637641
 ] 

ASF subversion and git services commented on GEODE-5787:
--------------------------------------------------------

Commit 4734957e6bdeffbad18d3d2828eec800353b25ef in geode's branch 
refs/heads/feature/GEODE-5787-dunit-internal from [~apa...@the9muses.net]
[ https://gitbox.apache.org/repos/asf?p=geode.git;h=4734957 ]

GEODE-5787: supporting bounce on windows

Signed-off-by: Sai Boorlagadda <sboorlaga...@pivotal.io>


> Support graceful bouncing DUnit VMs on windows
> ----------------------------------------------
>
>                 Key: GEODE-5787
>                 URL: https://issues.apache.org/jira/browse/GEODE-5787
>             Project: Geode
>          Issue Type: New Feature
>          Components: tests
>            Reporter: Sai Boorlagadda
>            Priority: Major
>              Labels: pull-request-available
>          Time Spent: 1h
>  Remaining Estimate: 0h
>
> Process.destroy is not graceful on windows and causes distributed system to 
> get into suspect processing and network partition when used in DUnit to 
> restart VMs.
> There were existing JDK 
> [bugs|https://bugs.openjdk.java.net/browse/JDK-8056139] open requesting for 
> supporting graceful destroy for Windows. 
> Underlying native implementation on Linux can take advantage of SIGTERM vs 
> SIGKILL and JDK support process.destroy and process.destroyForcibly, where as 
> on Windows this differentiation is not possible so its unlikely that we will 
> see JDK support for graceful destroy.
> A better approach would be to invoke 'System.exit( n )' using VM.invoke.



--
This message was sent by Atlassian JIRA
(v7.6.3#76005)

Reply via email to