flink-issues mailing list archives

Site index · List index
Message view « Date » · « Thread »
Top « Date » · « Thread »
From "ASF GitHub Bot (JIRA)" <j...@apache.org>
Subject [jira] [Commented] (FLINK-7580) Let LeaderGatewayRetriever implementations automatically retry failed gateway retrieval operations
Date Mon, 18 Sep 2017 13:20:00 GMT

    [ https://issues.apache.org/jira/browse/FLINK-7580?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16169999#comment-16169999
] 

ASF GitHub Bot commented on FLINK-7580:
---------------------------------------

Github user tillrohrmann commented on the issue:

    https://github.com/apache/flink/pull/4643
  
    Thanks for the review @zentol. Do I assume correctly that you also approved #4602 with
your review? 
    
    Rebasing onto the latest master. If Travis gives green light, I would like to merge this
PR.


> Let LeaderGatewayRetriever implementations automatically retry failed gateway retrieval
operations
> --------------------------------------------------------------------------------------------------
>
>                 Key: FLINK-7580
>                 URL: https://issues.apache.org/jira/browse/FLINK-7580
>             Project: Flink
>          Issue Type: Improvement
>          Components: Distributed Coordination, Webfrontend
>    Affects Versions: 1.4.0
>            Reporter: Till Rohrmann
>            Assignee: Till Rohrmann
>            Priority: Minor
>
> The {{LeaderGatewayRetrieval}} implementations {{AkkaJobManagerRetriever}} and the {{RpcGatewayRetriever}}
should automatically retry failed gateway retrieval operations. This could be the case if
the {{WebRuntimeMonitor}} is started before the actual Akka/RPC component. I would propose
to retry it a fixed number of times with a short delay in between. If the resolution fails
after exceeding the retries, a new retrieval operation will be started when requesting information
from the {{WebRuntimeMonitor}} with FLINK-7533. This ensures that the retry operation won't
run forever but also that it will eventually connect to the Akka/RPC component if it is existent.



--
This message was sent by Atlassian JIRA
(v6.4.14#64029)

Mime
View raw message