Tim, What you are looking for is not currently the case, the external shutdown behavior is different from the one initiated by the input operator. I created a JIRA to fix this a while ago:
https://issues.apache.org/jira/browse/APEXCORE-294 Thomas On Tue, Jan 26, 2016 at 2:54 PM, Amol Kekre <[email protected]> wrote: > I think the original design for shutdown was to start killing from input > operator side and have it go through the DAG. Though, I have not looked at > that in a while. Thomas would know the details. > > Thks, > Amol > > > On Tue, Jan 26, 2016 at 2:40 PM, Timothy Farkas <[email protected]> > wrote: > > > Hello all, > > > > If an Apex app is Shutdown (as opposed to killed) is there a guarantee > that > > all the operators will reach the last window produced by an input > operator > > and that committed will be called for that last window before the > > application is terminated? > > > > Thanks, > > Tim > > >
