flink-user mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From Gary Yao <g...@data-artisans.com>
Subject Re: Testing on Flink 1.5
Date Tue, 24 Apr 2018 07:29:11 GMT
Hi Amit,

web.timeout should only affect RPC calls originating from the REST API. In
FLIP-6, the submission of the job graph happens via HTTP. The value under
akka.ask.timeout is still used as the default timeout for RPC calls [1][2].
Since you also had custom heartbeats settings, you should consider setting
heartbeat.interval and heartbeat.timeout when using FLIP-6 mode [3]. AFAIK
Akka's DeathWatch is not used anymore to detect TaskManager failures. Hence,
akka.watch.heartbeat.interval should have no effect.

Best,
Gary

[1]
https://github.com/apache/flink/blob/fb254763c00df5d336c6defa1ae960e32c97b2ae/flink-runtime/src/main/java/org/apache/flink/runtime/entrypoint/ClusterEntrypoint.java#L389
[2]
https://github.com/apache/flink/blob/fb254763c00df5d336c6defa1ae960e32c97b2ae/flink-runtime/src/main/scala/org/apache/flink/runtime/akka/AkkaUtils.scala#L606
[3]
https://ci.apache.org/projects/flink/flink-docs-master/ops/config.html#heartbeat-manager

On Fri, Apr 20, 2018 at 12:22 PM, Amit Jain <aj2011it@gmail.com> wrote:

> Hi Gary,
>
> This setting has resolved the issue. Does it increase timeout for all the
> RPC or specific components?
>
> We had following settings in Flink 1.3.2 and they did the job for us.
>
> akka.watch.heartbeat.pause: 600 s
> akka.client.timeout: 5 min
> akka.ask.timeout: 120 s
>
>
> --
> Thanks,
> Amit
>

Mime
View raw message