commons-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "Sebb (JIRA)" <j...@apache.org>
Subject [jira] [Updated] (EXEC-99) Provide a non-blocking PumpStreamHandler
Date Sat, 09 Jan 2016 16:35:39 GMT

     [ https://issues.apache.org/jira/browse/EXEC-99?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel
]

Sebb updated EXEC-99:
---------------------
    Description: 
Sometimes the current PumpStreamHandler blocks when a process is killed but its child process
streams are processed by PumpStreamHandler

* The behaviour is not always reproducable and might depend on JDK and/or OS  
* Closing the processed stream does not seem work 
* Using Thread.interrupt() on reading from a stream does not work due to an old JDK issue
* Thread.stop() is deprecated and has no effect on JDK 8
* Hence the the stream pumper thread is stuck and can't terminate
* So we hang in PumpStreamHandler().stop

There was a bug fix to avoid hanging forever but this leaves the pumper thread behind which
could break a production system.

As a work-around I propose a "NonBlockingPumpStreamHandler" 

* use only on worker thread to process a streams
* using Thread.sleep() between processing of the streams
* can be stopped safely since there are no blocking operations
* it will be much slower processing the streams which could block the child process

It might be a good idea to hide those implementation behind a factory



  was:
Sometimes the current PumpStreamHandler when a process is killed but its child process streams
are processed by PumpStreamHandler

* The behaviour is not always reproducable and might depend on JDK and/or OS  
* Closing the processed stream does not seem work 
* Using Thread.interrupt() on reading from a stream does not work due to an old JDK issue
* Thread.stop() is deprecated and has no effect on JDK 8
* Hence the the stream pumper thread is stuck and can't terminate
* So we hang in PumpStreamHandler().stop

There was a bug fix to avoid hanging forever but this leaves the pumper thread behind which
could break a production system.

As a work-around I propose a "NonBlockingPumpStreamHandler" 

* use only on worker thread to process a streams
* using Thread.sleep() between processing of the streams
* can be stopped safely since there are no blocking operations
* it will be much slower processing the streams which could block the child process

It might be a good idea to hide those implementation behind a factory




> Provide a non-blocking PumpStreamHandler
> ----------------------------------------
>
>                 Key: EXEC-99
>                 URL: https://issues.apache.org/jira/browse/EXEC-99
>             Project: Commons Exec
>          Issue Type: Improvement
>    Affects Versions: 1.3
>            Reporter: Siegfried Goeschl
>            Assignee: Siegfried Goeschl
>             Fix For: 1.4
>
>
> Sometimes the current PumpStreamHandler blocks when a process is killed but its child
process streams are processed by PumpStreamHandler
> * The behaviour is not always reproducable and might depend on JDK and/or OS  
> * Closing the processed stream does not seem work 
> * Using Thread.interrupt() on reading from a stream does not work due to an old JDK issue
> * Thread.stop() is deprecated and has no effect on JDK 8
> * Hence the the stream pumper thread is stuck and can't terminate
> * So we hang in PumpStreamHandler().stop
> There was a bug fix to avoid hanging forever but this leaves the pumper thread behind
which could break a production system.
> As a work-around I propose a "NonBlockingPumpStreamHandler" 
> * use only on worker thread to process a streams
> * using Thread.sleep() between processing of the streams
> * can be stopped safely since there are no blocking operations
> * it will be much slower processing the streams which could block the child process
> It might be a good idea to hide those implementation behind a factory



--
This message was sent by Atlassian JIRA
(v6.3.4#6332)

Mime
View raw message