Kafka

Jun, thanks for your interest. We follow self-serve model. I just granted you permission to assign tickets. Side note: if you contribute, you might want to subscribe to the dev mailing list, which is more suitable for this purpose. -Matthias On 4/10/26 6:45 PM, 조형준 wrote: > Hi, > > I'd like to contribute to Apache Kafka. Could you please add me to the > Kafka contributors group on JIRA? I'm interested in working on KAFKA-20364 > > JIRA ID: xxxxxxjun > > Thanks, Jun >

Hello Kafka Community, We have a use case that requires attaching custom metadata to topics — things like SLA, durability guarantees, and PagerDuty account references. While exploring this in Kafka 4.2.0, we noticed an interesting inconsistency between broker-level and topic-level configuration behavior. Broker-level custom configs work fine: Running the following command succeeds without errors: ./bin/kafka-configs.sh --bootstrap-server localhost:9092 \ --entity-type brokers --entity-name 1 \ --alter --add-config 'durability.level=1' And describing the broker returns the config as expected: ./bin/kafka-configs.sh --bootstrap-server localhost:9092 \ --entity-type brokers --entity-name 1 \ --describe Output: Dynamic configs for broker 1 are: durability.level=null sensitive=true synonyms={DYNAMIC_BROKER_CONFIG:durability.level=null} Topic-level custom configs throw an exception: Running a similar command on a topic: ./bin/kafka-...

Hi Manikumar, The CVE seems to be still reserved and not published yet. Best, On Tue, Apr 7, 2026 at 1:49 PM Manikumar <manikumar@apache.org> wrote: > Severity: moderate > > Affected versions: > > - Apache Kafka Clients (org.apache.kafka:kafka-clients) 2.8.0 through 3.9.1 > - Apache Kafka Clients (org.apache.kafka:kafka-clients) 4.0.0 through 4.0.1 > - Apache Kafka Clients (org.apache.kafka:kafka-clients) 4.1.0 through 4.1.1 > > Description: > > A race condition in the Apache Kafka Java producer client’s buffer > pool management can cause messages to be silently delivered to > incorrect topics. > > When a produce batch expires due to delivery.timeout.ms while a > network request containing that batch is still in flight, the batch’s > ByteBuffer is prematurely deallocated and returned to the buffer pool. > If a subsequent producer batch—potentially destined for a different > topic—reuses this fr...

Severity: moderate Affected versions: - Apache Kafka Clients (org.apache.kafka:kafka-clients) 2.8.0 through 3.9.1 - Apache Kafka Clients (org.apache.kafka:kafka-clients) 4.0.0 through 4.0.1 - Apache Kafka Clients (org.apache.kafka:kafka-clients) 4.1.0 through 4.1.1 Description: A race condition in the Apache Kafka Java producer client’s buffer pool management can cause messages to be silently delivered to incorrect topics. When a produce batch expires due to delivery.timeout.ms while a network request containing that batch is still in flight, the batch’s ByteBuffer is prematurely deallocated and returned to the buffer pool. If a subsequent producer batch—potentially destined for a different topic—reuses this freed buffer before the original network request completes, the buffer contents may become corrupted. This can result in messages being delivered to unintended topics without any error being reported to the producer. Data Confidentiality: Messages in...

I am not familiar with the details. Sorry. But many people run with KRaft in production for years already, so I would believe that it's not something to be seriously concerned about. Also, looking into the Jira ticket, a lot of tasks are already resolved: - https://issues.apache.org/jira/browse/KAFKA-12466 (only one open task about "shell tool integration") - https://issues.apache.org/jira/browse/KAFKA-10310 (22/36 tickets resolved) -Matthias On 3/26/26 3:14 PM, Dima Brodsky via users wrote: > Hi Matthias, > > Question regarding 4.x and controller log compaction / snapshotting > facilities. As far as I understand KIP-630 > < https://cwiki.apache.org/confluence/display/KAFKA/KIP-630%3A+Kafka+Raft+Snapshot > > is > still open and being worked on. There is still no mechanism in kraft kafka > 4.x to compact and snapshot controller metadata logs. Is there any concern > that these logs will grow q...

Hello We are running Apache Kafka 3.9.0 in z/OS. We noticed following exception/warning in Kafka logs (it comes frequently, causing space issues due to the huge logs): Ý2026-03-26 23:15:56,003¨ WARN ÝSocketServer listenerType=ZK_BROKER, nodeId=1¨ Unexpected error from / 127.0.0.1 (channelId=127.0.0.1:9092-127.0.0.1:50290-10); closing connection (org.apache.kafka.common.network.Selector) org.apache.kafka.common.KafkaException: Size of FileRecords /global/kafka/SYS/kafka-logs/xxxx.infeed.private.v1-0/00000000000000000000.log has been truncated during write: old size 194, new size 97 .at org.apache.kafka.common.record.FileRecords.writeTo(FileRecords.java:298) .at org.apache.kafka.common.record.DefaultRecordsSend.writeTo(DefaultRecordsSend.java:34) .at org.apache.kafka.common.record.RecordsSend.writeTo(RecordsSend.java:50) .at org.apache.kafka.common.record.MultiRecordsSend.writeTo(MultiRecordsSend.java:93) .at org.apache.kafka.common.network.NetworkSend.write...

Hi Matthias, Question regarding 4.x and controller log compaction / snapshotting facilities. As far as I understand KIP-630 < https://cwiki.apache.org/confluence/display/KAFKA/KIP-630%3A+Kafka+Raft+Snapshot > is still open and being worked on. There is still no mechanism in kraft kafka 4.x to compact and snapshot controller metadata logs. Is there any concern that these logs will grow quickly enough in a large, busy, cluster to impact disk usage or controller startup times? Thanks! ttyl Dima On Thu, Mar 26, 2026 at 1:47 PM Matthias J. Sax < mjsax@apache.org > wrote: > Yes, we recently did 3.9.2 release, because the move to 4.x (which is > the first version w/o ZK) is a somewhat larger step for some users. > > At this point, it's unclear if there will be 3.9.3 release or not. But > as more time progresses, and we get new releases 4.3, 4.4 out, the > probability for a 3.9.3 release decreases. > > I would highly enco...