A work around might be to set the parallelism to 1 as Till suggested and
write into a network-shared directory.
Nonetheless, it sounds like a good feature to support strict assignments of
data sink tasks to machines.
Fabian
2015-05-22 16:10 GMT+02:00 Hilmi Yildirim :
> Hi Till,
> thank you very
Hi Till,
thank you very much.
BR,
Hilmi
Am 22.05.2015 um 16:09 schrieb Till Rohrmann:
Hi Hilmi,
in Flink you cannot control on which machines the individual tasks are
scheduled. Therefore, you cannot control on which machine the data is
written. However, you can control that only on one mac
Hi Hilmi,
in Flink you cannot control on which machines the individual tasks are
scheduled. Therefore, you cannot control on which machine the data is
written. However, you can control that only on one machine a file is
created by setting the degree of parallelism of the DataSink to *1*.
Cheers,
Hi,
I want to write the result of a batch process into a file. I want that
the result is written on a specific machine. For example, I have 4
machines (m1.example.com, m2.example.com, m3.example.com,
m4.example.com). Now, I want that the results go to m1 and that m1
writes the result in a file