[
https://issues.apache.org/jira/browse/FLINK-2993?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=15000134#comment-15000134
]
Ufuk Celebi commented on FLINK-2993:
------------------------------------
I agree that 100 seconds default is too high. Let's go for something lower.
But bear in mind that it will give a strange user experience when the
restarting does not work, because of missing task slots. If it is too
aggressive, you will see redeployment failures.
We should go for a less aggressive delay as default for HA (for now), where job
recovery is delayed by this timeout as well and the task managers have to be
notified about the new leader and reconnect within this time.
It will also be helpful to print this delay at the client when the state
changes to RESTARTING. What do you think?
> Set default DelayBetweenExecutionRetries to 0
> ---------------------------------------------
>
> Key: FLINK-2993
> URL: https://issues.apache.org/jira/browse/FLINK-2993
> Project: Flink
> Issue Type: Improvement
> Components: Core
> Affects Versions: 0.10
> Reporter: Stephan Ewen
> Assignee: Stephan Ewen
> Fix For: 1.0
>
>
> The default value is too high and gives a strange user experience.
--
This message was sent by Atlassian JIRA
(v6.3.4#6332)
