[
https://issues.apache.org/jira/browse/KAFKA-13152?page=com.atlassian.jira.plugin.system.issuetabpanels:comment-tabpanel&focusedCommentId=17712307#comment-17712307
]
Mickael Maison commented on KAFKA-13152:
----------------------------------------
KIP Freeze denotes the deadline when a KIP must be voted in order to be
considered in the next release. I typically don't see this deadline as a hard
limit but rather as a way to limit bringing large changes at the last minute.
KIP-770 was voted in 2021 so it's fine.
Feature freeze is when we stop merging PRs from KIPs or that introduce
significant changes. The goal is to stabilize the code base before the release
and have a couple of weeks to focus on fixing remaining bugs and stability
issues (like flaky tests).
It's not really holding up the release process yet. I've currently marked
KIP-770 for 3.6.0 but if Guozhang/Matthias think this would be a useful
addition for 3.5 and review and merge the PR in the next few days, I'm ok
including it in 3.5.0.
> Replace "buffered.records.per.partition" & "cache.max.bytes.buffering" with
> "{statestore.cache}/{input.buffer}.max.bytes"
> -------------------------------------------------------------------------------------------------------------------------
>
> Key: KAFKA-13152
> URL: https://issues.apache.org/jira/browse/KAFKA-13152
> Project: Kafka
> Issue Type: Improvement
> Components: streams
> Reporter: Guozhang Wang
> Assignee: Sagar Rao
> Priority: Major
> Labels: kip
> Fix For: 3.6.0
>
>
> The current config "buffered.records.per.partition" controls how many records
> in maximum to bookkeep, and hence it is exceed we would pause fetching from
> this partition. However this config has two issues:
> * It's a per-partition config, so the total memory consumed is dependent on
> the dynamic number of partitions assigned.
> * Record size could vary from case to case.
> And hence it's hard to bound the memory usage for this buffering. We should
> consider deprecating that config with a global, e.g. "input.buffer.max.bytes"
> which controls how much bytes in total is allowed to be buffered. This is
> doable since we buffer the raw records in <byte[], byte[]>.
--
This message was sent by Atlassian Jira
(v8.20.10#820010)
