This 100th edition of the Kafka Monthly Digest covers what happened in the Apache Kafka community in May 2026.
It's hard to believe this series has been going for 100 editions. Since the first edition the overall format has not changed very much. I just added the community releases section in early 2020. In the first edition (no longer available online) we looked at the Kafka 1.0.1 release as well as KIP-247 to KIP-264!
Releases
There are two new releases, 4.3.0 and 4.2.1:
4.3.0
The vote on 4.3.0 RC2 passed on May 20 and I released Kafka 4.3.0 on May 23. You can find the announcement on the Apache Kafka blog. You can also check the release notes and the release plan in the wiki.
With 25 KIPs implemented this release brings several major new features and many improvements across all components.
Kafka brokers and clients
Updates to the Kafka broker and clients include the following:
- New configuration to bootstrap new followers from the last tiered offsets instead of log start offsets. (KIP-1023)
Kafka Connect
Updates to Kafka Connect include the following:
Kafka Streams
Updates to Kafka Streams include the following:
4.2.1
After 5 release candidates, PoAn Yang released 4.2.1 on June 2. This bug fix release fixes 26 issues. You can find the announcement on the blog and the release plan in the wiki.
Kafka Improvement Proposals
Last month, the community submitted 21 KIPs (KIP-1333 to KIP-1355, KIP-1353 and 1354 were skipped). I'll highlight a few of them:
- KIP-1333: Support log segments larger than 2 GB: Today, the log segment size is stored in Kafka as an
intso log segments are limited to ~2GB in size. In very high throughput or long retention scenarios this limit can be problematic. This KIP proposes removing that limit allowing sizes up to Long.MAX_VALUE.
- KIP-1338: Structured MDC Context for Kafka Logging: This KIP proposes including Mapped Diagnostic Context (MDC) data into the logs emitted by all the Kafka components. This would make it easier for log analysis tools to parse and process Kafka logs.
- KIP-1347: Overriding voter set on storage formatting: In dynamic KRaft quorum mode, the hostnames of the voters are persisted in the metadata log and this is what controllers use at startup to discover the current voters. One limitation is that if all DNS names change at the same time, for example this can happen in Kubernetes environments when renaming a DNS domain, then controllers can't bootstrap any longer as all the hostnames in the metadata log are now invalid. This KIP aims at resolving this issue by providing a tool to update the voters in the metadata log.
- KIP-1352: Ranges and Range Aggregations in the Kafka Streams DSL: This KIP proposes adding event-centric aggregations, instead of usual time-centric, by basing them on record count and time (before and after an anchor) to the Streams DSL API.
Community Releases
I selected releases of some open source community projects:
- kroxylicious 0.21.0: Kroxylicious is an open source pluggable framework for writing network proxies that understand the Apache Kafka protocol. The new release adds support for automatic sidecar injection to transparently proxy traffic from applications. It also adds support for HAProxy PROXY Protocol and brings improvements to schema validation and AWS KMS support.
- Sarama 1.50: Sarama is a pure Golang Kafka client. This release brings a partitioner that is compatible with the Java client. It also updates the protocol version supported for group and offset fetch APIs.
Blogs
I selected some interesting blog articles that were published last month: