ebsite is still in 2023, or can I only compile from the
> source code?
>
>
> 在 2024年11月27日，17:56，William Wallace 写道：
>
>
> hi,
> It seems similar to issue described here:
> https://lists.apache.org/thread/g8yb4rlj0mlf1vgjl71815nts8r1w51p
> were we were not able to rest
hi,
It seems similar to issue described here:
https://lists.apache.org/thread/g8yb4rlj0mlf1vgjl71815nts8r1w51p
were we were not able to restore state because of the high number of S3
reads (in your case it might first encounter the connection limitation
first).
Have a look at https://issues.apache.
r to us.
>>
> Copy from the PR:
> Flink state restore from S3 is super slow because skip function is
> consuming ~15 seconds for ~6Mb of data.
> ...
> In this PR the skip going to be called only in case of compression because
> otherwise a stream is seekable.
>
> G
to cherry-pick this PR [1] at top of your Flink
> distro when possible.
> Additionally turn off state compression. These should do the trick...
>
> [1] https://github.com/apache/flink/pull/25509
>
> G
>
>
> On Tue, Oct 15, 2024 at 1:03 PM William Wallace <
>
y tested the PR with load but if you can report back it would
> be helpful.
>
> [1] https://issues.apache.org/jira/browse/FLINK-36530
>
> BR,
> G
>
>
> On Fri, Aug 16, 2024 at 11:25 AM William Wallace <
> theanonymous31...@gmail.com> wrote:
>
>> Context
>>
Context
We have recently upgraded from Flink 1.13.6 to Flink 1.19. We consume data
from ~ 40k Kafka topic partitions in some environments. We are using
aligned checkpoints. We set state.storage.fs.memory-threshold: 500kb.
Problem
At the point when the state for operator using
topic-partition-off
