[ 
https://issues.apache.org/jira/browse/SAMZA-2057?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17034802#comment-17034802
 ] 

Xinyu Liu commented on SAMZA-2057:
----------------------------------

I am not sure this is the right behavior to close a processing-time window when 
entering the end-of-stream. If the window is every 1 min and the app shuts down 
at 30 sec, I don't think it's always correct to trigger the window and emit the 
pane. Processing timer should be triggered when the timer actually fires 
according to wallclock time.

If this is a desired behavior, it should be controlled by a flag saying 
closeWindowOnEndOfStream or something. i wouldn't call this is a bug.

> Window operator on encountering end of stream does not emit the last window 
> pane
> --------------------------------------------------------------------------------
>
>                 Key: SAMZA-2057
>                 URL: https://issues.apache.org/jira/browse/SAMZA-2057
>             Project: Samza
>          Issue Type: Bug
>            Reporter: Sanil Jain
>            Priority: Minor
>
> On encountering EndOfStreamMessage a shutdown sequence is triggered, there is 
> an existing bug in High-Level Api where the messages in the operators are not 
> flushed on end of the stream to the next operators. This results in missing 
> the messages while testing the Samza code, which makes assertions fail. A 
> simple way to reproduce is any writing a Test using Test Framework for High 
> Level Api job using Session Window or KeyedTumblingWindows



--
This message was sent by Atlassian Jira
(v8.3.4#803005)

Reply via email to