Hi Diogo, I saw similar issues already. The root cause is always users actually not using any Flink specific stuff, but going to the Parquet Writer of Hadoop directly. As you can see in your stacktrace, there is not one reference to any Flink class.
The solution usually is to use the respective Flink sink instead of bypassing them [1]. If you opt to implement it manually nonetheless, it's probably easier to bundle Hadoop from a non-Flink dependency. [1] https://ci.apache.org/projects/flink/flink-docs-release-1.10/dev/connectors/streamfile_sink.html On Thu, Apr 16, 2020 at 5:36 PM Diogo Santos <diogodssan...@gmail.com> wrote: > Hi Till, > > definitely seems to be a strange issue. The first time the job is loaded > (with a clean instance of the Cluster) the job goes well, but if it is > canceled or started again the issue came. > > I built an example here https://github.com/congd123/flink-s3-example > > You can generate the artifact of the Flink Job and start the cluster with > the configuration on the docker-compose. > > Thanks for helping > > > > > > > > -- > Sent from: > http://apache-flink-user-mailing-list-archive.2336050.n4.nabble.com/ > -- Arvid Heise | Senior Java Developer <https://www.ververica.com/> Follow us @VervericaData -- Join Flink Forward <https://flink-forward.org/> - The Apache Flink Conference Stream Processing | Event Driven | Real Time -- Ververica GmbH | Invalidenstrasse 115, 10115 Berlin, Germany -- Ververica GmbH Registered at Amtsgericht Charlottenburg: HRB 158244 B Managing Directors: Timothy Alexander Steinert, Yip Park Tung Jason, Ji (Toni) Cheng
