[
https://issues.apache.org/jira/browse/FLINK-5068?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15666620#comment-15666620
]
Ufuk Celebi commented on FLINK-5068:
------------------------------------
Do you have a restart strategy configured? Shouldn't this be recovered
eventually when the TM is up and the job restarts? Feel free to send me the
logs to uce at apache.
> YARN HA: Job scheduled before TM is up leading to "Not enough free slots"
> error
> -------------------------------------------------------------------------------
>
> Key: FLINK-5068
> URL: https://issues.apache.org/jira/browse/FLINK-5068
> Project: Flink
> Issue Type: Bug
> Components: YARN
> Affects Versions: 1.1.3
> Reporter: Gyula Fora
>
> On one occasion after a job manager failure, the job was could not be
> recovered as it was scheduled before the TM was up causing an unrecoverable
> failure.
> So I ended up with a running yarn jm + tm with enough slots with a job failed
> (and not restarting)
> Please drop me a mail for the logs :)
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
