flink-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Chet Masterson <chet.master...@yandex.com>
Subject Queryable State
Date Mon, 24 Apr 2017 11:45:52 GMT
<div>I moved up from running queryable state on a standalone Flink instance to a several
node cluster. My queries don't seem to be responding when I execute them on the cluster. A
few questions:</div><div> </div><div>1. The error I am getting:</div><div><div>WARN
[ReliableDeliverySupervisor] Association with remote system [akka.tcp://flink@x.x.x.x:6123]
has failed, address is now gated for [5000] ms. Reason: [Association failed with [akka.tcp://flink@x.x.x.x:6123]]
Caused by: [Connection refused: /x.x.x.x:6123]</div><div>2017/04/23 20:19:01.016
ERROR Actor not found for: ActorSelection[Anchor(akka.tcp://flink@x.x.x.x:6123/), Path(/user/jobmanager)]</div><div> </div><div>I
assume this is because Flink is not servicing requests on port :6123. I am using the default
RPC ports defined in flink-conf.yaml. I confirm nothing is listening on port 6123 by using
netstat on the flink nodes.</div><div> </div><div>2. I make the following
changes on all nodes to flink-conf.yaml, then restart the cluster</div><div> </div><div><div>#jobmanager.rpc.port:
6123</div><div>query.server.port: 6123</div><div>query.server.enable:
true</div><div><div> </div><div>3. Now port 6123 is open, as
viewed from netstat.</div><div> </div><div>My question: what is the
proper configuration for servicing external queries when running in a cluster? Can I use jobmanager.rpc.port:
6123 which works standalone, do I have to add query.server.port and query.server.enable? Which
port should I be using?</div><div> </div><div> </div></div></div></div>
View raw message