Yes, you need overrides .getScheduler() method.
*createOrGetFIFOScheduler** - *When you want make your paragraphs run one
by one
*createOrGetParallelScheduler - *When you want to allow run paragraphs
parallel. Your interprete() method need to be thread safe.
*createOrGetRemoteScheduler* - Which is for internal use. so you don't want
to use it in your implementation
Thanks,
moon
On Mon, Jun 8, 2015 at 3:32 AM Victor Manuel Garcia <victor.gar...@beeva.com>
wrote:
> Hi moon,
>
> is that I'm developing an interpreter but wanted it to be in parallel
> ... I think I found it where I need to change
>
> is here...?
>
> if (concurrentSQL()) {
> int maxConcurrency = 10;
> return SchedulerFactory.singleton().createOrGetParallelScheduler(
> SparkSqlInterpreter.class.getName() + this.hashCode(),
> maxConcurrency);
> }
>
> another question, what is the difference between:
> - createOrGetParallelScheduler
> - createOrGetFIFOScheduler
> and
> - createOrGetRemoteScheduler
>
> thks
>
> 2015-06-08 12:20 GMT+02:00 moon soo Lee <m...@apache.org>:
> > Hi,
> >
> > That's depends on implementation of interpreter.
> > For example, SparkSqlInterpreter (conditionally) allows parallel
> execution
> > of paragraphs, while SparkInterpreter doesn't.
> > Which do you use?
> >
> > Thanks,
> > moon
> >
> >
> > On Mon, Jun 8, 2015 at 2:19 AM Victor Manuel Garcia
> > <victor.gar...@beeva.com> wrote:
> >>
> >> Hi guys,
> >>
> >> I have a question, always being executed one paragraph expect to
> >> finish to run another ....?
> >> It is possible to avoid that...?
> >>
> >> thks
> >> --
> >> Victor Manuel García
> >> Ingeniero de Software
> >>
> >> [+34] 672104297 | victor.gar...@beeva.com
>
>
>
> --
> Victor Manuel García
> Ingeniero de Software
>
> [+34] 672104297 | victor.gar...@beeva.com
>
