flink-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Aljoscha Krettek <aljos...@apache.org>
Subject Re: accessing flink HA cluster with scala shell/zeppelin notebook
Date Tue, 24 Jan 2017 16:13:43 GMT
+Till Rohrmann <trohrmann@apache.org>, do you know what can be used to
access a HA cluster from that setting.

Adding Till since he probably knows the HA stuff best.

On Sun, 22 Jan 2017 at 15:58 Maciek Próchniak <mpr@touk.pl> wrote:

> Hi,
>
> I have standalone Flink cluster configured with HA setting (i.e. with
> zookeeper recovery). How should I access it remotely, e.g. with Zeppelin
> notebook or scala shell?
>
> There are settings for host/port, but with HA setting they are not fixed
> - if I check which is *current leader* host and port and set that I get
> exception on job manager:
>
> 20:36:38.237 [flink-akka.actor.default-dispatcher-22704] WARN
> o.a.f.runtime.jobmanager.JobManager - Discard message
> LeaderSessionMessage(null,SubmitJob(JobGraph(jobId:
> 02a3a43464b2a750e04855d4c0b6fecb),EXECUTION_RESULT_AND_STATE_CHANGES))
> because the expected leader session ID
> Some(1a4e9d39-2c59-45bb-b81c-d867bec1958d) did not equal the received
> leader session ID None.
>
> - I guess it's reasonable behaviour, since I should use appropriate
> LeaderRetrievalService and so on. But apparently there's no such
> possibility in scala flink shell?
>
> Is it missing feature? I can prepare patch, but I'm not sure how would I
> hook behaviour of ClusterClient into FlinkILoop?
>
> thanks,
>
> maciek
>
>

Mime
View raw message