[ 
https://issues.apache.org/jira/browse/NIFI-16603?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
 ]

David Handermann resolved NIFI-16603.
-------------------------------------
    Fix Version/s: 2.13.0
       Resolution: Fixed

> Bootstrap stop command reports failure after successful forced termination
> --------------------------------------------------------------------------
>
>                 Key: NIFI-16603
>                 URL: https://issues.apache.org/jira/browse/NIFI-16603
>             Project: Apache NiFi
>          Issue Type: Bug
>          Components: Core Framework
>            Reporter: Arvind Kandpal
>            Assignee: Anshu Singh
>            Priority: Minor
>             Fix For: 2.13.0
>
>          Time Spent: 1h 10m
>  Remaining Estimate: 0h
>
> When the application process does not stop within the graceful shutdown 
> timeout, StopBootstrapCommand falls back to ProcessHandle.destroyForcibly(). 
> The check on the return value of destroyForcibly() is inverted.
> When destroyForcibly() returns true (the kill request was accepted), the 
> command logs "Application Process [pid] force termination failed" and does 
> not wait for the process to exit, so the command status stays ERROR. The 
> branch that waits for exit and sets SUCCESS only runs when destroyForcibly() 
> returns false.
> Steps to reproduce:
> 1. Run StopBootstrapCommand against an application process that does not exit 
> on SIGTERM within the graceful shutdown timeout. For example, use a process 
> that ignores SIGTERM: sh -c  "trap '' TERM; while true; do sleep 1; done"
> 2. Run the stop command.
> Expected: the process is killed, the log says "force termination completed", 
> and the command status is SUCCESS. 
> Actual: the process is killed, but the log says "force termination failed" 
> and the command status is ERROR (exit code 1).
> This was introduced in NIFI-13665. The destroy() method a few lines above 
> handles its return value the right way.
> Code: nifi-bootstrap, StopBootstrapCommand.java, destroyForcibly()
> StopBootstrapCommandTest does not cover this case, because the mocked 
> destroyForcibly() returns false by default.



--
This message was sent by Atlassian Jira
(v10.3.26#10030026)

Reply via email to