Checkpoint was declined because one input stream is finished flink version 1.12.7

[Prasanna kumar <pr...@gmail.com>]

Hi,

We have the following job where one input stream will be finished earlier.
Because of this, the checkpointing is not happening causing the error.
What should be done to not stop checkpointing.

Source Read from Kafka apply broadcast rules and then write output to
kafka.



[image: Screen Shot 2022-01-10 at 1.12.22 PM.png]


[image: Screen Shot 2022-01-10 at 12.54.14 PM.png]

Thanks,
Prasanna.

Re: Checkpoint was declined because one input stream is finished flink version 1.12.7

[Piotr Nowojski <pn...@apache.org>]

Hi,

Yes, this is a well known limitation of older Flink versions. To make it
work you need to upgrade to Flink 1.14.x [1]. As of 1.14.x you would also
need to manually enable this feature [2]. As of 1.15.x (not yet released),
this feature will be enabled by default [3].

Best, Piotrek

[1] https://issues.apache.org/jira/browse/FLINK-2491
[2]
https://nightlies.apache.org/flink/flink-docs-master/docs/dev/datastream/fault-tolerance/checkpointing/#checkpointing-with-parts-of-the-graph-finished
[3] https://issues.apache.org/jira/browse/FLINK-25105

pon., 10 sty 2022 o 08:57 Prasanna kumar <pr...@gmail.com>
napisał(a):

> Hi,
>
> We have the following job where one input stream will be finished earlier.
> Because of this, the checkpointing is not happening causing the error.
> What should be done to not stop checkpointing.
>
> Source Read from Kafka apply broadcast rules and then write output to
> kafka.
>
>
>
> [image: Screen Shot 2022-01-10 at 1.12.22 PM.png]
>
>
> [image: Screen Shot 2022-01-10 at 12.54.14 PM.png]
>
> Thanks,
> Prasanna.
>
>