[
https://issues.apache.org/jira/browse/FLINK-4152?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15384069#comment-15384069
]
ASF GitHub Bot commented on FLINK-4152:
---------------------------------------
Github user tillrohrmann commented on the issue:
https://github.com/apache/flink/pull/2257
I think by imposing this contract we actually introduced problems we didn't
have before. Thus, in order to remedy these problems for the upcoming release
and go back a bit in the direction of the pre-RM age, I loosened the contract
of the RM so that it can no longer reject TM registrations. This makes sense in
my opinion, since we don't have a mean to shut down orphaned TMs anyway.
So in this version, the RM's task is to ensure that at least a predefined
set of resources is allocated and to notify the JM about a TM death (not
strictly mandatory).
What we could actually do is to also register orphaned TMs (or ones
registered by a different RM). Then we wouldn't have the problem that we
allocate too many resources for a JM.
> TaskManager registration exponential backoff doesn't work
> ---------------------------------------------------------
>
> Key: FLINK-4152
> URL: https://issues.apache.org/jira/browse/FLINK-4152
> Project: Flink
> Issue Type: Bug
> Components: Distributed Coordination, TaskManager, YARN Client
> Reporter: Robert Metzger
> Assignee: Till Rohrmann
> Attachments: logs.tgz
>
>
> While testing Flink 1.1 I've found that the TaskManagers are logging many
> messages when registering at the JobManager.
> This is the log file:
> https://gist.github.com/rmetzger/0cebe0419cdef4507b1e8a42e33ef294
> Its logging more than 3000 messages in less than a minute. I don't think that
> this is the expected behavior.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
