The only nodes that matter are those on which the Flink processes, i.e
Job- and TaskManagers
<https://ci.apache.org/projects/flink/flink-docs-release-1.4/concepts/runtime.html#job-managers-task-managers-clients>,
are being run.
To prevent a JobManager node failure from causing the job to fail you
have to look into an HA setup
<https://ci.apache.org/projects/flink/flink-docs-release-1.4/ops/jobmanager_high_availability.html>.
(The jobmanager is responsible for distributing/coordinating work)
In case of a TaskManager node failure the job will fail and restart
<https://ci.apache.org/projects/flink/flink-docs-release-1.4/dev/restart_strategies.html>,
provided that enough TaskManagers are
still alive to satisfy the resource requirements of the job.
Can you elaborate a bit more what happened when you used the --detached
param?
On 28.09.2017 16:33, r. r. wrote:
Thank you, Chesnay
to make sure - should the node where the job has been submitted goes down, the
processing will continue, I hope?
Do I need to ensure this by configuration?
btw I added --detached param to the run cmd, but it didn't go into background
process as I would've expected. Am I guessing wrong?
Thanks!
Rob
>-------- Оригинално писмо --------
>От: Chesnay Schepler ches...@apache.org
>Относно: Re: how many 'run -c' commands to start?
>До: user@flink.apache.org
>Изпратено на: 28.09.2017 15:05
Hi!
Given a Flink cluster, you would only call `flink run ...` to submit a
job once; for simplicity i would submit it on the node where you started
the cluster. Flink will automatically distribute job across the cluster,
in smaller independent parts known as Tasks.
Regards,
Chesnay
On 28.09.2017 08:31, r. r. wrote:
Hello
I successfully ran a job with 'flink run -c', but this is for the local
setup.
How should i proceed with a cluster? Will flink automagically instantiate
the job on all servers - i hope i don't have to start 'flink run -c' on all
machines.
New to flink and bigdata, so sorry for the probably silly question
Thanks!
Rob
