You refer to the DataSet (batch) API, right?
In that case you can evaluate your condition in the program and fetch a
DataSet back to the client using List<X> myData = DataSet<X>.collect();
Based on the result of the collect() call you can define and execute a new
program.
Note: collect() will immediately trigger the execution of the program in
its current state and bring the result back to the client. There is also a
size limitation of results that can be fetched back. This is the Akka
framesize which is 10MB by default but could be adapted.
It would look similar to this:
ExecutionEnvironment env = ...
DataSet<X> a = env.readFile(...);
List<Y> b = a.groupBy().reduce().filter().collect();
DataSet<Z> c;
if(b.get(0).equals(...)) {
c = env.readFile(someFile);
} else {
c = env.readFile(someOtherFile);
}
c.map().groupBy().reduce()....writeAsFile(result);
env.execute();
Cheers, Fabian
2015-10-30 22:40 GMT+01:00 Giacomo Licari <[email protected]>:
> Hi guys,
> I would ask to you how could I create triggers in Flink.
>
> I would like to perform some operations on a dataset and according to some
> conditions, based on an attribute of a Pojo class or Tuple, execute some
> triggers.
> I mean, starting collecting other datasources' data and performing
> operations over them.
>
> An Example.
> I have a dataset of Pojo class Person. My trigger activation condition is
> (number of italian people > 100).
> If so, I collect another datasource and I execute operations over it.
>
> Do you think is that possible in Flink?
>
> Thanks,
> Giacomo
>
