* Fri Oct 02 2026 Dirk Müller <dmueller@suse.com>
- update to 2.15.1:
* `DeserializingConsumer` and `DeserializingShareConsumer` now
deserialize the message key before the value, matching
`SerializingProducer` and the Java client. When both key and
value fail to deserialize, the key error is now
surfaced instead of the value error.
* Minor fix for subjectPrefix parameter in subjects API
* Fix race conditions
* Fix segmentation fault after calling
`AdminClient.delete_records()` followed
* by another Admin API call (e.g. `list_topics()`) on Python
3.14.
* Use `asyncio.get_running_loop()` instead of
`asyncio.get_event_loop()` to avoid creating a new event loop
and raise an error in case a loop isn't available
(@AlexCai26, #2339).
* confluent-kafka-python 2.15.1 is based on librdkafka 2.15.1,
see the librdkafka release notes for a complete list of changes,
enhancements, fixes and upgrade considerations.
- update to 2.15.0 (bsc#1284013, CVE-2026-15911):
* confluent-kafka-python 2.15.0 adds a **Preview**
implementation of the KIP-932 share consumer (Queues for
Kafka). Members of a share group cooperatively consume from
the same partitions with per-record acquire/acknowledge
semantics and redelivery, providing queue-like consumption on
top of Kafka.
* New `ShareConsumer` and `DeserializingShareConsumer` clients:
subscribe, batch poll (`poll()` returns a `Messages` batch),
and close (with context-manager support).
* Implicit (default) and explicit acknowledgement modes
(`share.acknowledgement.mode`), with `AcknowledgeType` ACCEPT
/ RELEASE / REJECT and `Message.delivery_count()`.
* Synchronous and asynchronous commit (`commit_sync` /
`commit_async`) and an acknowledgement-commit callback.
* Runtime SASL credential updates (`set_sasl_credentials`) and
new `IllegalStateException` /
`ConcurrentModificationException` exceptions.
* Fix Encryption fails when expanded union types have two
references to the same record (@ChristophMajcenAtXxxlutz,
[#2262])
* Make orjson optional in JSON serdes with stdlib json fallback
* Handle non-http errors during retries
* Use TLS certification verification with Hashicorp Vault
* confluent-kafka-python v2.15.0 is based on librdkafka
v2.15.0, see the
* librdkafka release notes
* for a complete list of changes, enhancements, fixes and
upgrade considerations.
- update to 2.14.2:
* Fix URL joining in Python client
* Handle anyOf/allOf in JSON transforms
* Fix redefining a named Avro type in a diamond dependency
pattern
* Fix typing_extensions import errors on Python 3.12
* Fix Encryption fails when referencing self defined types
* confluent-kafka-python v2.14.2 is based on librdkafka
v2.14.2, see the librdkafka release notes for a complete list
of changes, enhancements, fixes and upgrade considerations.
* Sat Apr 11 2026 Dirk Müller <dmueller@suse.com>
- update to 2.14.0:
* Implement async context manager protocol for AIOProducer and
AIOConsumer
* Add AssociatedNameStrategy
* Add enableAt to RuleSet
* Ensure normalize.schemas config is passed during Protobuf ref
lookup #2214
* Fix type annotations for context manager hooks so that they
are correct for subclasses
* Fix OAuth callback handling for Async IO clients to prevent
initialization failures
* confluent-kafka-python v2.14.0 is based on librdkafka
v2.14.0, see the
* librdkafka release notes
* for a complete list of changes, enhancements, fixes and
upgrade considerations.
- update to 2.13.2:
* v2.13.2 is a maintenance release with the following fixes and
enhancements:
* Add Confluent-Client-Version header to requests to SR, Python
client
* Add UAMI OAuth changes
* Fixed memory leak in Producer.produce() when called with
headers and raises BufferError (queue full) or RuntimeError
(producer closed). The allocated rd_headers memory is now
properly freed in error paths before returning. Fixes Issue
https://github.com/confluentinc/confluent-kafka-
python/issues/2167.
* Fixed type hinting of KafkaError class, Consumer.__init()__,
Producer.__init()__, Producer.produce() and Consumer.commit()
and introduced a script in tools directory to keep error
codes up to date. Fixes Issue
https://github.com/confluentinc/confluent-kafka-
python/issues/2168.
* Fix the token expiration logic in SR Oauth
* Ensure use of cachetools is thread-safe
* Remove passing resolver in json validate
- update to 2.13.0:
* Enforced type hinting for all interfaces
* Handle OAuth Token Refreshes using background thread for
Admin, Producer and Consumer clients
* Added black and isort linting rules and enforcement to
codebase
* Enabled direct creation of Message objects
* Added `close()` method to producer
* Added context manager for librdkafka classes to enable easy
scope cleanup
* Expose deterministic partitioner functions
* Add Accept-Version header for schemas
* Enhanced the BufferTimeoutManager to flush the librdkafka
queue
* Remove experimental module designation for Async classes
* Add `__len__` function to AIOProducer
* Enhance Message class to include serialisation support and
rich comparison
* Support Strict Validation Flags in Avro Serializers
* Type hint `__enter__` to return the same object type that
called it
* Fixed `Consumer.poll()`, `Consumer.consume()`,
`Producer.poll()`, and `Producer.flush()` blocking
indefinitely and not responding to Ctrl+C (KeyboardInterrupt)
signals. The implementation now uses a "wakeable poll"
pattern that breaks long blocking calls into smaller chunks
(200ms) and periodically re-acquires the Python GIL to check
for pending signals. This allows Ctrl+C to properly interrupt
blocking operations. Fixes Issues #209 and #807 (#2126).
* Fix support for wrapped Avro unions
* Fixed segfault exceptions on calls against objects that had
closed internal objects
* Handle evolution during field transformation of schemas
* Handle null group name to prevent segfault in Admin
`list_consumer_group_offsets()`
* Ensure schemaId initialization is thread-safe
* Fix error propagation rule for Python's C API
* Fix SR delete behavior with client-side caching
* Don't leave NewTopic partially-initialized on error
* Fix formatting issue of TopicPartition that causes
SystemError on Windows
* Add Py_None check for msg and offset params in `commit()` and
`store_offsets()`
* Fix experimental module references
* Update outdated example in README.md
* Fri Apr 03 2026 Dirk Müller <dmueller@suse.com>
- update to 2.12.1:
* Restored macOS binaries compatibility with macOS 13
* `libversion()` now returns the string/integer tuple for
varients on version -- use `version()` for string only
response
* Added Python 3.14 support and dropped 3.7 support -- Free-
threaded capabilities not fully supported yet
* Fixed use.schema.id in `sr.lookup_schema()`
* Removed tomli dependency from standard (non-documentation)
requirements
* Fixed experimental asyncio example files to correctly use new
capabilities
* Fixed invalid argument error on schema lookups on repeat
requests
* Fixed documentation generation and added error checks for
builds to prevent future breaks
- update to 2.12.0:
* Starting with __confluent-kafka-python 2.12.0__, the next
generation consumer group rebalance protocol defined in
* *KIP-848** is **production-ready**. Please refer to the
following migration guide for moving from `classic` to
`consumer` protocol.
* **Note:** The new consumer group protocol defined in KIP-848
is not enabled by default. There are few contract change
associated with the new protocol and might cause breaking
changes. `group.protocol` configuration property dictates
whether to use the new `consumer` protocol or older `classic`
protocol. It defaults to `classic` if not provided.
* ### AsyncIO Producer (experimental)
* Introduces beta class `AIOProducer` for asynchronous message
production in asyncio applications.
* AsyncIO Producer (experimental): Introduces beta class
`AIOProducer` for
* asynchronous message production in asyncio applications. This
API offloads
* blocking librdkafka calls to a thread pool and schedules
common callbacks
* (`error_cb`, `throttle_cb`, `stats_cb`, `oauth_cb`, `logger`)
onto the event
* loop for safe usage inside async frameworks.
* #### Features
* Batched async produce: `await AIOProducer(...).produce(topic,
value=...)`
* buffers messages and flushes when the buffer threshold or
timeout is reached.
* Async lifecycle: `await producer.flush()`, `await
producer.purge()`, and
* transactional operations (`init_transactions`,
`begin_transaction`,
* `commit_transaction`, `abort_transaction`).
- update to 2.10.1:
* Handled `None` value for optional `ctx` parameter in
`ProtobufDeserializer`
* Handled `None` value for optional `ctx` parameter in
`AvroDeserializer`
- update to 2.10.0:
* [KIP-848] Group Config is now supported in AlterConfigs,
IncrementalAlterConfigs and DescribeConfigs.
* [KIP-848] `describe_consumer_groups()` now supports KIP-848
introduced `consumer` groups. Two new fields for consumer
group type and target assignment have also been added. Type
defines whether this group is a `classic` or `consumer`
group. Target assignment is only valid for the `consumer`
protocol and its defaults to NULL. (#1873).
* Fri Jul 18 2025 Dirk Müller <dmueller@suse.com>
- update to 2.8.0:
* Ensure algorithm query param is passed for CSFLE
* DGS-19492 Handle records nested in arrays/maps when searching
for tags
* confluent-kafka-python v2.8.0 is based on librdkafka v2.8.0,
see the librdkafka release notes
- update to 2.7.0:
* Support for Data Contracts with Schema Registry, including
+ Data Quality rules
+ Data Transformation rules
+ Client-Side Field Level Encryption (CSFLE)
+ Schema Migration rules (requires Python 3.9+)
* Migrated the Schema Registry client from requests to httpx
* Add support for multiple URLs
* Allow configuring timeout
* Fix deletion semantics
* Python deserializer can take SR client
* Fix handling of Avro unions
* Remove deprecated RefResolver for JSON
* Support delete of subject version
* Added missing dependency on googleapis-common-protos when
using protobufs.
- update to 2.6.2:
* Support for Data Contracts with Schema Registry, including
* Data Quality rules
* Data Transformation rules
* Client-Side Field Level Encryption (CSFLE)
* Schema Migration rules (requires Python 3.9+)
* Migrated the Schema Registry client from requests to httpx
* Add support for multiple URLs
* Allow configuring timeout
* Fix deletion semantics
* Python deserializer can take SR client
* Fix handling of Avro unions
* Remove deprecated RefResolver for JSON
* Support delete of subject version
* Added Python 3.13 wheels (#1828).
* KIP-460 Admin Leader Election RPC
* Added Transactional id resource type for ACL operations
- update to 2.5.3:
* Fix an assert being triggered during push telemetry call when
no metrics matched on the client side.
* Minor enhancement to JSONDeserializer to retrieve schema from
Schema Registry if not provided
* confluent-kafka-python is based on librdkafka v2.5.3, see the
* librdkafka release notes
* for a complete list of changes, enhancements, fixes and
upgrade considerations.
- update to 2.5.0:
* KIP-107 Added delete_records API.
* Added an example to show the usage of the custom logger with
`AdminClient`.
* Improve caching on Schema Registry client.
* Removed usage of `strcpy` to enhance security of the client.
* Removed support for centos6 and centos7.
* Fixed invalid write in `OAUTHBEARER/OIDC` extensions copy.
* Fixed documentation for default value of `operation_timeout`
and `request_timeout` in various Admin apis.
* Fixed an issue related to import error of `TopicCollection`
and `TopicPartitionInfo` classes when importing through other
module like mypy.
* Fixed a segfault when `commit` or `store_offsets` consumer
method is called incorrectly with errored Message object.
* Fixed `logger` not working when provided as an argument to
`AdminClient` instead of a configuration property.
* Fixed some memory leaks related to `PyDict_SetItem`.
- update to 2.4.0:
* KIP-848: Added KIP-848 based new consumer group rebalance
protocol. The feature is an **Early Access**: not production
ready yet. Please refer detailed doc for more information.
* Fix segfault with describe_topics and flaky connection
(@lpsinger, #1692)
* Fri May 23 2025 Markéta Machová <mmachova@suse.com>
- Convert to pip-based build