Hi Ventura!
I hope you can get along without editing the Flink runtime, with the help
of overriding the open/close methods of the RichFunctions and do your
initialization there.
They are called once per life of a parallel function (except in iterations,
where they are called once per superstep, bu
Hi Stephan!
Thank you for your reply, first of all! You're right about how I
distributed my data. I need this because I have an object that should be
shared among tasks. I am working on decoupling this object from the cuda
type at the moment and I will follow your suggestions!
About my CudaExecut
Hi Ventura!
You are distributing your data via something like "env.fromElements(...)"
or "env.fromCollection(...)", is that correct?
The master node (JobManager) currently takes each InputFormat and checks
whether it needs some "master side initialization". For file input formats,
this computes f
I am using Flink 0.9-SNAPSHOT, this is the complete stack trace:
org.apache.flink.client.program.ProgramInvocationException: The program
execution failed: Cannot initialize task 'DataSource (at
(DownpourSDG.java:28)
(org.apache.flink.api.java.io.CollectionInputFormat))': Deserializing the
InputFor
Hi Ventura!
Flink loads on startup only the Flink classes and all user code classes are
loaded dynamically. Each point when a user class is used, or a user class
object is deserialized, Flink uses the user code classloader of that
specific job.
For you as the user, this has the implication that i
The corresponding code snippet could also help.
Cheers,
Till
On Wed, Apr 22, 2015 at 4:45 PM, Robert Metzger wrote:
> Hi,
>
> which version of Flink are you using?
>
> Can you send us the complete stack trace of the error to help us
> understand the exact location where the issue occurs?
>
> O
Hi,
which version of Flink are you using?
Can you send us the complete stack trace of the error to help us understand
the exact location where the issue occurs?
On Wed, Apr 22, 2015 at 4:33 PM, Ventura Del Monte <
venturadelmo...@gmail.com> wrote:
> Hello, I am working on a flink-based deep lea
Hello, I am working on a flink-based deep learning library for my master's
thesis. I am experiencing this issue at the moment: I have a java class
with a transient field, so I had to write both a kryo custom serializer and
a java one. The (de)serialization needs to access another object of my
syste
