[
https://issues.apache.org/jira/browse/FLINK-32895?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17763394#comment-17763394
]
Rui Fan commented on FLINK-32895:
---------------------------------
Hi [~zhuzh] , I created the FLIP-364 in advance due to I found several points
in the restart strategy that need to be improved. We can discuss them in the
mail list in the future.
There are 2 option for discussion:
* Option1: Start discuss FLIP-364 after deprecating the RestartStrategies is
discussed.
* Option2: FLIP-364 has serveral points need to be discussed, we can first
discuss other parts of FLIP-364 besides RestartStrategies. And the
RestartStrategies part can be followed by your separate FLIP.
WDYT?
BTW, after some more thought:
restart-strategy.exponential-delay.fail-on-exceeding-max-backoff may not work
well. Because the user may want to restart this job multiple times using
max-backoff before failing it.
For example, users don't want the delay-time to be too long, so they set the
initial-backoff=1s, backoff-multiplier=2, max-backoff=30s. So the delay time is
1s, 2s, 4s, 8s, 16s, 30s, 30s, 30s, 30s, 30s, etc. If we introduced the
`fail-on-exceeding-max-backoff`, it means that the job won't restart when the
delay-time is extended to 30s at first time. right?
Please correct me if I'm wrong, and looking forward to more feedbacks from
community, thanks~
[1]https://cwiki.apache.org/confluence/display/FLINK/FLIP-364%3A+Improve+the+restart-strategy
> Introduce the max attempts for Exponential Delay Restart Strategy
> -----------------------------------------------------------------
>
> Key: FLINK-32895
> URL: https://issues.apache.org/jira/browse/FLINK-32895
> Project: Flink
> Issue Type: Improvement
> Components: Runtime / Coordination
> Reporter: Rui Fan
> Assignee: Rui Fan
> Priority: Major
> Labels: pull-request-available
>
> Currently, Flink has 3 restart strategies, they are: fixed-delay,
> failure-rate and exponential-delay.
> The exponential-delay is suitable if a job continues to fail for a period of
> time. The fixed-delay and failure-rate has the max attempts mechanism, that
> means, the job won't restart and go to fail after the attempt exceeds the
> threshold of max attempts.
> The max attempts mechanism is reasonable, flink should not or need to
> infinitely restart the job if the job keeps failing. However, the
> exponential-delay doesn't have the max attempts mechanism.
> I propose introducing the
> `restart-strategy.exponential-delay.max-attempts-before-reset` to support the
> max attempts mechanism for exponential-delay. It means flink won't restart
> job if the number of job failures before reset exceeds
> max-attempts-before-reset when is exponential-delay is enabled.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
