[
https://issues.apache.org/jira/browse/FLINK-10748?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=16719118#comment-16719118
]
Till Rohrmann commented on FLINK-10748:
---------------------------------------
Thanks for reporting this issue [~jeroens]. I think you are right that the
current setup is not ideal. I think by now, the {{RestServerEndpoint}} no
longer needs to be colocated with the {{Dispatcher}} and the {{JobMaster}} as
it was the case before. Thus, it should be possible to simply use the retrieved
leader gateway in the {{RedirectHandler}} and give it to the requested handler
(also if this means that the communication with the {{Dispatcher}} is a remote
call) instead of redirecting the client to another {{RestServerEndpoint}}.
> Jobmanager in HA setup, redirects (307) don't work when behind a load balancer
> ------------------------------------------------------------------------------
>
> Key: FLINK-10748
> URL: https://issues.apache.org/jira/browse/FLINK-10748
> Project: Flink
> Issue Type: Bug
> Components: JobManager
> Affects Versions: 1.3.3, 1.4.2, 1.5.5, 1.6.2
> Reporter: Jeroen Steggink
> Priority: Critical
>
> In a HA Jobmanager setup, connecting to a follower results in a redirect
> (HTTP/1.1 307 Temporary Redirect) to the leader. However, it redirects to an
> ip address (why is it not the hostname?) which is in another network and not
> reachable. I have configured hostnames, not ip addresses.
> Wouldn't it be better to proxy the request to another jobmanager instead of
> using a redirect?
>
--
This message was sent by Atlassian JIRA
(v7.6.3#76005)
