Using existing flink cluster with zeppelin

[Jens Grassel <je...@wegtam.com>]

Hi,

we try to use an existing flink cluster with zeppelin following the
official documentation at:

http://zeppelin.apache.org/docs/0.8.0-SNAPSHOT/setup/deployment/flink_and_spark_cluster.html

We are using current zeppelin compiled from source with flink 1.3.2
support and a flink 1.3.2 cluster.

However, after following the documentation it seems that zeppelin more
or less starts its own flink node and does not connect to the
existing cluster. Jobs are executed fine and logged in the interpreter
log file. But they do not show up in the flink cluster web ui (running &
completed jobs).

Is there anything we might be missing?
According to the flink documentation a cluster must be configured using
a yaml file. But there is none in the zeppelin directory.

Regards,

Jens

-- 
CTO, Wegtam GmbH, 09. Nebelung 2017, 13:59
Homepage : https://www.wegtam.com

Ours is a world of nuclear giants and ethical infants.
		-- General Omar N. Bradley

Re: Using existing flink cluster with zeppelin

[Jens Grassel <je...@wegtam.com>]

Hi,

On Thu, 9 Nov 2017 14:05:42 +0100
Jens Grassel <je...@wegtam.com> wrote:

JG> However, after following the documentation it seems that zeppelin
JG> more or less starts its own flink node and does not connect to the
JG> existing cluster. Jobs are executed fine and logged in the
JG> interpreter log file. But they do not show up in the flink cluster
JG> web ui (running & completed jobs).

as it seems not all jobs are listed in the flink job manager...
Some jobs are listed in the job/task manager. So it seems that they are
actually executed in the cluster.

Has anyone experienced such issues before?

Regards,

Jens

-- 
CTO, Wegtam GmbH, 09. Nebelung 2017, 14:31
Homepage : https://www.wegtam.com

A city is a large community where people are lonesome together.
		-- Herbert Prochnow