v4.0.4

2025-04-09

Library updates

• [fix][sec] Upgrade jwt/v5 to 5.2.2 to address CVE-2025-30204 (#24140)
• [fix][sec] Upgrade pulsar-function-go dependencies to address CVE-2025-22870 (#24135)
• [improve] Upgrade Netty to 4.1.119.Final (#24049)

Broker

• [fix] Avoid negative estimated entry count (#24055)
• [fix][broker] Add expire check for replicator (#23975)
• [fix][broker] Avoid IllegalStateException when marker_type field is not set in publishing (#24087)
• [fix][broker] fix delay queue sequence issue. (#24035)
• [fix][broker] Fix Metadata Event Synchronizer producer creation retry so that the producer gets created eventually (#24081)
• [fix][broker] Fix Metadata event synchronizer should not fail with bad version (#24080)
• [fix][broker] Fix missing validation when setting retention policy on topic level (#24032)
• [fix][broker] Fix NPE while publishing Metadata-Event with not init producer (#24079)
• [fix][broker] Fix UnsupportedOperationException while setting subscription level dispatch rate policy (#24048)
• [fix][broker] http metric endpoint get compaction latency stats always be 0 (#24067)
• [fix][broker] Pattern subscription doesn't work when the pattern excludes the topic domain. (#24072)
• [fix][broker] Restore the behavior to dispatch batch messages according to consumer permits (#24092)
• [fix][broker] topics infinitely failed to delete after remove cluster from replicated clusters modifying when using partitioned system topic (#24097)
• [fix][broker]Fix failed consumption after loaded up a terminated topic (#24063)
• [fix][ml] Corrected pulsar_storage_size metric to not multiply offloaded storage by the write quorum (#24054)
• [fix][ml] Don't estimate number of entries when ledgers are empty, return 1 instead (#24125)
• [fix][ml] Fix issues in estimateEntryCountBySize (#24089)
• [fix][ml] Return 1 when bytes size is 0 or negative for entry count estimation (#24131)
• [improve][broker] Change topic exists log to warn (#24116)
• [improve][broker] extract getMaxEntriesInThisBatch into a method and add unit test for it (#24117)
• [improve][broker] Optimize message expiration rate repeated update issues (#24073)
• [improve][broker] Optimize ThresholdShedder with improved boundary checks and parameter reuse (#24064)
• [improve][broker][branch-4.0] PIP-406: Introduce metrics related to dispatch throttled events (#24111)
• [improve][meta] Change log level from error to warn for unknown notification types in OxiaMetadataStore (#24126)

Client

• [fix][client] Copy eventTime to retry letter topic and DLQ messages (#24059)
• [fix][client] Fix building broken batched message when publishing (#24061)
• [fix][client] Fix consumer leak when thread is interrupted before subscribe completes (#24100)
• [fix][client] Pattern subscription regression when broker-side evaluation is disabled (#24104)
• [fix][client] PIP-409 retry/dead letter topic producer config don't take effect. (#24071)
• [improve][client] PIP-409: support producer configuration for retry/dead letter topic producer (#24020)
• [improve][client] Prevent NullPointException when closing ClientCredentialsFlow (#24123)
• [clean][client] Clean code for the construction of retry/dead letter topic name (#24082)

Pulsar IO and Pulsar Functions

• [fix][io] Fix KinesisSink json flattening for AVRO's SchemaType.BYTES (#24132)
• [improve][fn] Introduce NewOutputMessageWithError to enable error handling (#24122)
• [improve][io] Enhance Kafka connector logging with focused bootstrap server information (#24128)
• [improve][io] Remove sleep when sourceTask.poll of kafka return null (#24124)

Others

• [fix][doc] fix doc related to chunk message feature. (#24023)
• [fix][doc] Workaround Go Yaml issue go-yaml/yaml#789 in docker-compose example (#24040)
• [improve][monitor] Add version=0.0.4 to /metrics content type for Prometheus 3.x compatibility (#24060)

Tests & CI

• [fix][ci] Bump dependency-check to 12.1.0 to fix OWASP Dependency Check job (#24083)
• [fix][test] Fix flaky NonPersistentTopicTest.testMsgDropStat (#24134)
• [fix][test] Fix flaky PrometheusMetricsTest.testBrokerMetrics (#24042)

For the complete list, check the full changelog.

v3.3.6

2025-04-09

Library updates

• [fix] Bump google.golang.org/protobuf from 1.32.0 to 1.33.0 in /pulsar-function-go (#22261)
• [fix][sec] Upgrade jwt/v5 to 5.2.2 to address CVE-2025-30204 (#24140)
• [fix][sec] Upgrade pulsar-function-go dependencies to address CVE-2025-22870 (#24135)
• [improve] Upgrade Netty to 4.1.119.Final (#24049)

Broker

• [fix] Avoid negative estimated entry count (#24055)
• [fix][broker] Add expire check for replicator (#23975)
• [fix][broker] Avoid IllegalStateException when marker_type field is not set in publishing (#24087)
• [fix][broker] Fix Metadata Event Synchronizer producer creation retry so that the producer gets created eventually (#24081)
• [fix][broker] Fix Metadata event synchronizer should not fail with bad version (#24080)
• [fix][broker] Fix missing validation when setting retention policy on topic level (#24032)
• [fix][broker] Fix NPE while publishing Metadata-Event with not init producer (#24079)
• [fix][broker] Pattern subscription doesn't work when the pattern excludes the topic domain. (#24072)
• [fix][broker] topics infinitely failed to delete after remove cluster from replicated clusters modifying when using partitioned system topic (#24097)
• [fix][broker]Fix failed consumption after loaded up a terminated topic (#24063)
• [fix][ml] Corrected pulsar_storage_size metric to not multiply offloaded storage by the write quorum (#24054)
• [fix][ml] Don't estimate number of entries when ledgers are empty, return 1 instead (#24125)
• [fix][ml] Fix issues in estimateEntryCountBySize (#24089)
• [fix][ml] Return 1 when bytes size is 0 or negative for entry count estimation (#24131)
• [improve][broker] Change topic exists log to warn (#24116)
• [improve][broker] Optimize message expiration rate repeated update issues (#24073)
• [improve][broker] Optimize ThresholdShedder with improved boundary checks and parameter reuse (#24064)
• [improve][broker] Separate offload read and write thread pool (#24025)
• [improve][meta] Change log level from error to warn for unknown notification types in OxiaMetadataStore (#24126)

Client

• [fix][client] Copy eventTime to retry letter topic and DLQ messages (#24059)
• [fix][client] Fix building broken batched message when publishing (#24061)
• [fix][client] Fix consumer leak when thread is interrupted before subscribe completes (#24100)
• [fix][client] Pattern subscription regression when broker-side evaluation is disabled (#24104)
• [improve][client] Prevent NullPointException when closing ClientCredentialsFlow (#24123)
• [clean][client] Clean code for the construction of retry/dead letter topic name (#24082)

Pulsar IO and Pulsar Functions

• [fix][io] Fix KinesisSink json flattening for AVRO's SchemaType.BYTES (#24132)
• [improve][io] Enhance Kafka connector logging with focused bootstrap server information (#24128)
• [improve][io] Remove sleep when sourceTask.poll of kafka return null (#24124)

Others

• [fix][doc] fix doc related to chunk message feature. (#24023)
• [improve][monitor] Add version=0.0.4 to /metrics content type for Prometheus 3.x compatibility (#24060)

Tests & CI

• [fix][ci] Bump dependency-check to 12.1.0 to fix OWASP Dependency Check job (#24083)
• [fix][test] Fix flaky NonPersistentTopicTest.testMsgDropStat (#24134)
• [fix][test] Fix flaky PrometheusMetricsTest.testBrokerMetrics (#24042)

For the complete list, check the full changelog.

v3.0.11

2025-04-09

Library updates

• [fix] Bump google.golang.org/protobuf from 1.32.0 to 1.33.0 in /pulsar-function-go (#22261)
• [fix][sec] Upgrade jwt/v5 to 5.2.2 to address CVE-2025-30204 (#24140)
• [fix][sec] Upgrade pulsar-function-go dependencies to address CVE-2025-22870 (#24135)
• [improve] Upgrade Netty to 4.1.119.Final (#24049)

Broker

• [fix] Avoid negative estimated entry count (#24055)
• [fix][admin] Verify is policies read only before revoke permissions on topic (#23730)
• [fix][broker] Add expire check for replicator (#23975)
• [fix][broker] Avoid IllegalStateException when marker_type field is not set in publishing (#24087)
• [fix][broker] fix broker identifying incorrect stuck topic (#24006)
• [fix][broker] Fix BucketDelayedDeliveryTracker thread safety (#24014)
• [fix][broker] Fix get outdated compactedTopicContext after compactionHorizon has been updated (#20984)
• [fix][broker] Fix Metadata Event Synchronizer producer creation retry so that the producer gets created eventually (#24081)
• [fix][broker] Fix Metadata event synchronizer should not fail with bad version (#24080)
• [fix][broker] Fix NPE while publishing Metadata-Event with not init producer (#24079)
• [fix][broker] Fix UnsupportedOperationException while setting subscription level dispatch rate policy (#24048)
• [fix][broker] Geo Replication lost messages or frequently fails due to Deduplication is not appropriate for Geo-Replication (#23697)
• [fix][broker] http metric endpoint get compaction latency stats always be 0 (#24067)
• [fix][broker] Pattern subscription doesn't work when the pattern excludes the topic domain. (#24072)
• [fix][broker] topics infinitely failed to delete after remove cluster from replicated clusters modifying when using partitioned system topic (#24097)
• [fix][broker]Fix failed consumption after loaded up a terminated topic (#24063)
• [fix][ml] Corrected pulsar_storage_size metric to not multiply offloaded storage by the write quorum (#24054)
• [fix][ml] Don't estimate number of entries when ledgers are empty, return 1 instead (#24125)
• [fix][ml] Fix issues in estimateEntryCountBySize (#24089)
• [fix][ml] Return 1 when bytes size is 0 or negative for entry count estimation (#24131)
• [improve][broker] Change topic exists log to warn (#24116)
• [improve][broker] Fix non-persistent system topic schema compatibility (#23286)
• [improve][broker] Improve CompactedTopicImpl lock (#20697)
• [improve][broker] Make the estimated entry size more accurate (#23931)
• [improve][broker] Optimize message expiration rate repeated update issues (#24073)
• [improve][broker] Optimize ThresholdShedder with improved boundary checks and parameter reuse (#24064)
• [improve][ml] Use lock-free queue in InflightReadsLimiter since there's no concurrent access (#23962)

Client

• [fix][client] Copy eventTime to retry letter topic and DLQ messages (#24059)
• [fix][client] Fix building broken batched message when publishing (#24061)
• [fix][client] Fix consumer leak when thread is interrupted before subscribe completes (#24100)
• [fix][client] Pattern subscription regression when broker-side evaluation is disabled (#24104)
• [improve][client] Prevent NullPointException when closing ClientCredentialsFlow (#24123)
• [clean][client] Clean code for the construction of retry/dead letter topic name (#24082)

Pulsar IO and Pulsar Functions

• [fix][io] Fix KinesisSink json flattening for AVRO's SchemaType.BYTES (#24132)
• [improve][fn] Set default tenant and namespace for ListFunctions cmd (#23881)
• [improve][io] Enhance Kafka connector logging with focused bootstrap server information (#24128)
• [improve][io] Remove sleep when sourceTask.poll of kafka return null (#24124)

Others

• [fix][doc] fix doc related to chunk message feature. (#24023)
• [improve][cli] Support additional msg metadata for V1 topic on peek message cmd (#23978)
• [improve][monitor] Add version=0.0.4 to /metrics content type for Prometheus 3.x compatibility (#24060)

Tests & CI

• [fix][test] Fix flaky test OneWayReplicatorUsingGlobalZKTest.testConfigReplicationStartAt (#24011)
• [fix][test]Fix flaky test V1_ProducerConsumerTest.testConcurrentConsumerReceiveWhileReconnect (#24019)
• [fix][ci] Bump dependency-check to 12.1.0 to fix OWASP Dependency Check job (#24083)
• [improve][test] Upgrade Testcontainers to 1.20.4 and docker-java to 3.4.0 (#24003)
• [improve][ci] Upgrade Gradle Develocity Maven Extension to 1.23.1 (#24004)

For the complete list, check the full changelog.

v4.0.3

2025-02-27

Known issues

There's a potential regression where consuming messages could fail due to #23931 changes. This is being addressed in #24055 and #24089. No workarounds currently exist.

Library updates

• [feat][misc] upgrade oxia java client to 0.5.0 (#23871)
• [fix] Bump org.apache.solr:solr-core from 8.11.3 to 9.8.0 in /pulsar-io/solr (#23899)
• [fix] Use Alpine 3.21 in base image (#23964)
• [fix][build] Upgrade json-smart to 2.5.2 (#23966)
• [improve] Upgrade to Netty 4.1.117.Final (#23863)
• [fix][sec] Upgrade to Netty 4.1.118 to address CVE-2025-24970 (#23965)

Broker

• [fix][broker] Fix bug in RangeCache where different instance of the key wouldn't ever match (#23903)
• [fix][broker] Apply dispatcherMaxReadSizeBytes also for replay reads for Shared and Key_Shared subscriptions (#23894)
• [fix][broker] Closed topics won't be removed from the cache (#23884)
• [fix][broker] Fix BucketDelayedDeliveryTracker thread safety (#24014)
• [fix][broker] Fix NPE during active consumer is disconnected (#24008)
• [fix][broker] Fix incorrect blockedConsumerOnUnackedMsgs value when maxUnackedMessagesPerConsumer is 1 (#23796)
• [fix][broker] Fix rate limiter token bucket and clock consistency issues causing excessive throttling and connection timeouts (#23930)
• [fix][broker] Fix repeatedly acquired pending reads quota (#23869)
• [fix][broker] Fix seeking by timestamp can be reset the cursor position to earliest (#23919)
• [fix][broker] Geo Replication lost messages or frequently fails due to Deduplication is not appropriate for Geo-Replication (#23697)
• [fix][broker] Make InflightReadsLimiter asynchronous and apply it for replay queue reads (#23901)
• [fix][broker] PIP-322 Fix issue with rate limiters where rates can exceed limits initially and consumption pauses until token balance is positive (#24012)
• [fix][broker] PIP-379 Key_Shared implementation race condition causing out-of-order message delivery (#23874)
• [fix][broker] Skip to persist cursor info if it failed by cursor closed (#23615)
• [fix][broker] fix broker identifying incorrect stuck topic (#24006)
• [fix][broker] fix broker may lost rack information (#23331)
• [fix][meta] Fix ephemeral Zookeeper put which creates a persistent znode (#23984)
• [fix][meta] Fix ephemeral handling of ZK nodes and fix MockZooKeeper ephemeral and ZK stat handling (#23988)
• [fix][ml] Fix cursor metadata compatability issue when switching the config unackedRangesOpenCacheSetEnabled (#23759)
• [fix][ml] Fix deadlock in PendingReadsManager (#23958)
• [fix][ml] Fix memory leak due to duplicated RangeCache value retain operations (#23955)
• [fix][ml] Fix memory leaks in ManagedCursorInfo and ManagedLedgerInfo decompression and compression (#23960)
• [fix][ml] incorrect non-durable cursor's backlog due to concurrently trimming ledger and non-durable cursor creation (#23951)
• [improve][broker] Avoid PersistentReplicator.expireMessages logic compute backlog twice (#23957)
• [improve][broker] Avoid logging errors when there is a connection issue during subscription. (#23939)
• [improve][broker] Avoid printing log for IncompatibleSchemaException in ServerCnx (#23938)
• [improve][broker] Do not print error logs for NotFound or Conflict errors when using the Admin API (#23928)
• [improve][broker] Don't print error logs for ProducerBusyException (#23929)
• [improve][broker] Fix non-persistent system topic schema compatibility (#23286)
• [improve][broker] Improve Consumer.equals performance (#23864)
• [improve][broker] Make the estimated entry size more accurate (#23931)
• [improve][broker] Refactor a private method to eliminate an unnecessary parameter (#23915)
• [improve][broker] Remove spamming logs for customized managed ledger (#23862)
• [improve][broker] Support values up to 2^32 in ConcurrentBitmapSortedLongPairSet (#23878)
• [improve][meta] Simplify getting parent path in ZKMetadataStore without using java.io.File (#23996)
• [improve][ml] Use lock-free queue in InflightReadsLimiter since there's no concurrent access (#23962)

Client

• [feat][client] Support forward proxy for the ZTS server in pulsar-client-auth-athenz (#23947)
• [fix] Avoid NPE when closing an uninitialized SameAuthParamsLookupAutoClusterFailover (#23911)
• [fix] Initialize UrlServiceProvider before trying to use transaction coordinator (#23914)
• [fix][client] Fix LoadManagerReport not found (#23886)
• [fix][client] Fix memory leak in ClientCnx.newLookup when there's TooManyRequestsException (#23971)
• [fix][client] Fix memory leak when message size exceeds max message size and batching is enabled (#23967)
• [fix][client] call redeliver 1 msg but did 2 msgs (#23943)
• [fix][client] fix retry topic with exclusive mode. (#23859)
• [improve][client] Avoid logging errors for retriable errors when creating producer (#23935)
• [improve][client] Update TypedMessageBuilder deliverAfter and deliverAt api comment (#23969)
• [improve][cli] Support additional msg metadata for V1 topic on peek message cmd (#23978)

Pulsar IO and Pulsar Functions

• [fix][io] Fix pulsar-io:pom not found (#23979)
• [improve][fn] Set default tenant and namespace for ListFunctions cmd (#23881)
• [improve][io] Allow skipping connector deployment (#23932)

Others

• [fix] fix for code scanning alert no. 48: Uncontrolled data used in path expression (#23985)
• [fix][build] Use amazoncorretto:21-alpine3.20 JDK build for Alpine 3.20 (#23898)
• [fix][build] Add develops for buildtools (#23992)
• [improve][proxy] Make keep-alive interval configurable in Pulsar Proxy (#23981)

Tests & CI

• [fix][test] Fix flaky DelayedDeliveryTest.testEnableTopicDelayedDelivery (#23893)
• [fix][test] Fix flaky test MetadataStoreTest.emptyStoreTest (#23998)
• [fix][test] Fix flaky test OneWayReplicatorUsingGlobalZKTest.testConfigReplicationStartAt (#24011)
• [fix][test] Fix quiet time implementation in BrokerTestUtil.receiveMessages (#23876)
• [fix][test] fix flaky testNegativeAcksWithBackoff when batch enabled. (#23986)
• [fix][test]Fix flaky test V1_ProducerConsumerTest.testConcurrentConsumerReceiveWhileReconnect (#24019)
• [improve][test] Add solution to PulsarMockBookKeeper for intercepting reads (#23875)
• [improve][test] Support decorating topic, subscription, dispatcher, ManagedLedger and ManagedCursors instances in tests (#23892)
• [improve][test] Upgrade Testcontainers to 1.20.4 and docker-java to 3.4.0 (#24003)
• [fix][ci] Configure Docker data-root to /mnt/doc...

v3.3.5

2025-02-27

Known issues

There's a potential regression where consuming messages could fail due to #23931 changes. This is being addressed in #24055 and #24089. No workarounds currently exist.

Library updates

• [feat][misc] upgrade oxia java client to 0.5.0 (#23871)
• [fix] Use Alpine 3.21 in base image (#23964)
• [fix][build] Upgrade json-smart to 2.5.2 (#23966)
• [improve] Upgrade to Netty 4.1.117.Final (#23863)
• [fix][sec] Upgrade to Netty 4.1.118 to address CVE-2025-24970 (#23965)

Broker

• [fix][broker] Fix bug in RangeCache where different instance of the key wouldn't ever match (#23903)
• [fix][broker] Closed topics won't be removed from the cache (#23884)
• [fix][broker] Continue using the next provider for http authentication if one fails (#23842)
• [fix][broker] Fix BucketDelayedDeliveryTracker thread safety (#24014)
• [fix][broker] Fix NPE during active consumer is disconnected (#24008)
• [fix][broker] Fix acknowledgeCumulativeAsync block when ackReceipt is enabled (#23841)
• [fix][broker] Fix incorrect blockedConsumerOnUnackedMsgs value when maxUnackedMessagesPerConsumer is 1 (#23796)
• [fix][broker] Fix possible mark delete NPE when batch index ack is enabled (#23833)
• [fix][broker] Fix rate limiter token bucket and clock consistency issues causing excessive throttling and connection timeouts (#23930)
• [fix][broker] Fix repeatedly acquired pending reads quota (#23869)
• [fix][broker] Fix the retry mechanism in MetadataCache#readModifyUpdateOrCreate (#23686)
• [fix][broker] Geo Replication lost messages or frequently fails due to Deduplication is not appropriate for Geo-Replication (#23697)
• [fix][broker] Make InflightReadsLimiter asynchronous and apply it for replay queue reads (#23901)
• [fix][broker] PIP-322 Fix issue with rate limiters where rates can exceed limits initially and consumption pauses until token balance is positive (#24012)
• [fix][broker] PIP-399: Fix Metric Name for Delayed Queue (#23712)
• [fix][broker] Remove blocking calls from internalGetPartitionedStats (#23832)
• [fix][broker] Revert "[fix][broker] Cancel possible pending replay read in cancelPendingRead (#23384)" (#23855)
• [fix][broker] Support large number of unack message store for cursor recovery (#9292)
• [fix][broker] fix broker identifying incorrect stuck topic (#24006)
• [fix][broker] fix broker may lost rack information (#23331)
• [fix][meta] Fix ephemeral Zookeeper put which creates a persistent znode (#23984)
• [fix][meta] Fix ephemeral handling of ZK nodes and fix MockZooKeeper ephemeral and ZK stat handling (#23988)
• [fix][ml] Fix cursor metadata compatability issue when switching the config unackedRangesOpenCacheSetEnabled (#23759)
• [fix][ml] Fix deadlock in PendingReadsManager (#23958)
• [fix][ml] Fix memory leak due to duplicated RangeCache value retain operations (#23955)
• [fix][ml] Fix memory leaks in ManagedCursorInfo and ManagedLedgerInfo decompression and compression (#23960)
• [fix][ml] incorrect non-durable cursor's backlog due to concurrently trimming ledger and non-durable cursor creation (#23951)
• [improve][broker] Avoid PersistentReplicator.expireMessages logic compute backlog twice (#23957)
• [improve][broker] Avoid logging errors when there is a connection issue during subscription. (#23939)
• [improve][broker] Avoid printing log for IncompatibleSchemaException in ServerCnx (#23938)
• [improve][broker] Do not print error logs for NotFound or Conflict errors when using the Admin API (#23928)
• [improve][broker] Don't print error logs for ProducerBusyException (#23929)
• [improve][broker] Fix non-persistent system topic schema compatibility (#23286)
• [improve][broker] Improve Consumer.equals performance (#23864)
• [improve][broker] Make the estimated entry size more accurate (#23931)
• [improve][broker] Reduce unnecessary REPLICATED_SUBSCRIPTION_SNAPSHOT_REQUEST (#23839)
• [improve][broker] Remove spamming logs for customized managed ledger (#23862)
• [improve][broker] Support values up to 2^32 in ConcurrentBitmapSortedLongPairSet (#23878)
• [improve][meta] Simplify getting parent path in ZKMetadataStore without using java.io.File (#23996)
• [improve][ml] Use lock-free queue in InflightReadsLimiter since there's no concurrent access (#23962)

Client

• [fix] Avoid NPE when closing an uninitialized SameAuthParamsLookupAutoClusterFailover (#23911)
• [fix] Initialize UrlServiceProvider before trying to use transaction coordinator (#23914)
• [fix][client] Fix LoadManagerReport not found (#23886)
• [fix][client] Fix memory leak in ClientCnx.newLookup when there's TooManyRequestsException (#23971)
• [fix][client] Fix memory leak when message size exceeds max message size and batching is enabled (#23967)
• [fix][client] Orphan producer when concurrently calling producer closing and reconnection (#23853)
• [fix][client] call redeliver 1 msg but did 2 msgs (#23943)
• [fix][client] fix retry topic with exclusive mode. (#23859)
• [improve][client] Avoid logging errors for retriable errors when creating producer (#23935)
• [improve][cli] Support additional msg metadata for V1 topic on peek message cmd (#23978)

Pulsar IO and Pulsar Functions

• [fix][io] Fix pulsar-io:pom not found (#23979)
• [improve] Validate user paths in Functions utils (#22833)
• [improve][fn] Set default tenant and namespace for ListFunctions cmd (#23881)
• [improve][io] Allow skipping connector deployment (#23932)

Others

• [fix] fix for code scanning alert no. 48: Uncontrolled data used in path expression (#23985)
• [fix][build] Use amazoncorretto:21-alpine3.20 JDK build for Alpine 3.20 (#23898)
• [fix][build] Add develops for buildtools (#23992)
• [fix][misc] Honor dynamic log levels in log4j2.yaml (#23847)
• [improve][proxy] Make keep-alive interval configurable in Pulsar Proxy (#23981)

Tests & CI

• [improve][test] Add solution to PulsarMockBookKeeper for intercepting reads (#23875)
• [improve][test] Upgrade Testcontainers to 1.20.4 and docker-java to 3.4.0 (#24003)
• [fix][test] Add reconsumeLater call in RetryTopicTest#testRetryTopicWithMultiTopic. (#23857)
• [fix][test] Fix flaky DelayedDeliveryTest.testEnableTopicDelayedDelivery (#23893)
• [fix][test] Fix flaky test MetadataStoreTest.emptyStoreTest (#23998)
• [fix][test] Fix flaky test OneWayReplicatorUsingGlobalZKTest.testConfigReplicationStartAt (#24011)
• [fix][test] Fix qui...

v3.0.10

2025-02-27

Library updates

• [improve] Upgrade to Netty 4.1.117.Final (#23863)
• [fix][sec] Upgrade to Netty 4.1.118 to address CVE-2025-24970 (#23965)
• [fix] Bump org.apache.solr:solr-core from 8.11.3 to 9.8.0 in /pulsar-io/solr (#23899)
• [fix][build] Upgrade json-smart to 2.5.2 (#23966)

Broker

• [fix][broker] Fix bug in RangeCache where different instance of the key wouldn't ever match (#23903)
• [fix][broker] Closed topics won't be removed from the cache (#23884)
• [fix][broker] Continue using the next provider for http authentication if one fails (#23842)
• [fix][broker] Fix acknowledgeCumulativeAsync block when ackReceipt is enabled (#23841)
• [fix][broker] Fix incorrect blockedConsumerOnUnackedMsgs value when maxUnackedMessagesPerConsumer is 1 (#23796)
• [fix][broker] Fix possible mark delete NPE when batch index ack is enabled (#23833)
• [fix][broker] Fix repeatedly acquired pending reads quota (#23869)
• [fix][broker] Fix the retry mechanism in MetadataCache#readModifyUpdateOrCreate (#23686)
• [fix][broker] Make InflightReadsLimiter asynchronous and apply it for replay queue reads (#23901)
• [fix][broker] PIP-399: Fix Metric Name for Delayed Queue (#23712)
• [fix][broker] Remove blocking calls from internalGetPartitionedStats (#23832)
• [fix][broker] Revert "[fix][broker] Cancel possible pending replay read in cancelPendingRead (#23384)" (#23855)
• [fix][broker] Support large number of unack message store for cursor recovery (#9292)
• [fix][broker] fix broker may lost rack information (#23331)
• [cleanup][admin] Do not print full stacktrace when get partitioned metadata not found (#20979)
• [fix][meta] Fix ephemeral Zookeeper put which creates a persistent znode (#23984)
• [fix][meta] Fix ephemeral handling of ZK nodes and fix MockZooKeeper ephemeral and ZK stat handling (#23988)
• [fix][misc] Honor dynamic log levels in log4j2.yaml (#23847)
• [fix][ml] Fix cursor metadata compatability issue when switching the config unackedRangesOpenCacheSetEnabled (#23759)
• [fix][ml] Fix deadlock in PendingReadsManager (#23958)
• [fix][ml] Fix memory leak due to duplicated RangeCache value retain operations (#23955)
• [fix][ml] Fix memory leaks in ManagedCursorInfo and ManagedLedgerInfo decompression and compression (#23960)
• [fix][ml] incorrect non-durable cursor's backlog due to concurrently trimming ledger and non-durable cursor creation (#23951)
• [improve][broker] Avoid PersistentReplicator.expireMessages logic compute backlog twice (#23957)
• [improve][broker] Avoid logging errors when there is a connection issue during subscription. (#23939)
• [improve][broker] Avoid printing log for IncompatibleSchemaException in ServerCnx (#23938)
• [improve][broker] Do not print error logs for NotFound or Conflict errors when using the Admin API (#23928)
• [improve][broker] Don't print error logs for ProducerBusyException (#23929)
• [improve][broker] Improve Consumer.equals performance (#23864)
• [improve][broker] Reduce unnecessary REPLICATED_SUBSCRIPTION_SNAPSHOT_REQUEST (#23839)
• [improve][broker] Remove spamming logs for customized managed ledger (#23862)
• [improve][broker] Support values up to 2^32 in ConcurrentBitmapSortedLongPairSet (#23878)
• [improve][meta] Simplify getting parent path in ZKMetadataStore without using java.io.File (#23996)

Client

• [feat][client] Support forward proxy for the ZTS server in pulsar-client-auth-athenz (#23947)
• [fix] Avoid NPE when closing an uninitialized SameAuthParamsLookupAutoClusterFailover (#23911)
• [fix] Initialize UrlServiceProvider before trying to use transaction coordinator (#23914)
• [fix][client] Fix LoadManagerReport not found (#23886)
• [fix][client] Fix memory leak in ClientCnx.newLookup when there's TooManyRequestsException (#23971)
• [fix][client] Fix memory leak when message size exceeds max message size and batching is enabled (#23967)
• [fix][client] Orphan producer when concurrently calling producer closing and reconnection (#23853)
• [fix][client] call redeliver 1 msg but did 2 msgs (#23943)
• [fix][client] fix retry topic with exclusive mode. (#23859)
• [improve][client] Avoid logging errors for retriable errors when creating producer #23935

Pulsar IO and Pulsar Functions

• [fix][io] Fix pulsar-io:pom not found (#23979)
• [improve] Validate user paths in Functions utils (#22833)
• [improve][io] Allow skipping connector deployment (#23932)

Others

• [improve][proxy] Make keep-alive interval configurable in Pulsar Proxy (#23981)
• [fix] fix for code scanning alert no. 48: Uncontrolled data used in path expression (#23985)
• [fix][build] Add develops for buildtools (#23992)
• [improve] Support overriding java.net.preferIPv4Stack with OPTS (#23846)

Tests & CI

• [improve][test] Add solution to PulsarMockBookKeeper for intercepting reads (#23875)
• [fix][test] Add reconsumeLater call in RetryTopicTest#testRetryTopicWithMultiTopic. (#23857)
• [fix][test] Fix flaky DelayedDeliveryTest.testEnableTopicDelayedDelivery (#23893)
• [fix][test] Fix quiet time implementation in BrokerTestUtil.receiveMessages (#23876)
• [fix][test] fix flaky testNegativeAcksWithBackoff when batch enabled. (#23986)
• [fix][test]Fix flaky test testTopicUnloadAfterSessionRebuild (#23852)
• [fix][ci] Configure Docker data-root to /mnt/docker to avoid running out of disk space (#23909)
• [improve][ci] Increase Maven max heap size to 2048M and tune GCLockerRetryAllocationCount (#23883)
• [improve][ci] Publish build scans to develocity.apache.org (#23851)
• [improve][ci] Skip "OWASP dependency check" when data wasn't found in cache (#23970)
• [improve][ci] Update GitHub Workflows to Ubuntu 22.04 (#20729)

For the complete list, check the full changelog.

v4.0.2

2025-01-20

Known Major Issues in this Release

This release contains a known issue impacting the seek method by timestamp. This issue was reported after the release was published.

• [Bug] Seeking by timestamp causes the subscription to reset to the earliest position (#23910)

Workaround for the issue:

The issue is caused by changes introduced in #22792. To resolve this, disable the functionality by setting the following configuration:

managedLedgerCursorResetLedgerCloseTimestampMaxClockSkewMillis=-1

Workaround Configuration Notes:

• This configuration key needs to be manually added to your broker.conf file as it doesn't exist by default
• For Apache Pulsar Helm Chart Kubernetes deployments, add the following to your values.yaml file under the broker/configData section:

broker:
  configData:
    PULSAR_PREFIX_managedLedgerCursorResetLedgerCloseTimestampMaxClockSkewMillis: "-1"

Library updates

• [fix][sec] Bump commons-io version to 2.18.0 (#23684)
• [fix][sec] Mitigate CVE-2024-53990 by disabling AsyncHttpClient CookieStore (#23725)
• [fix][sec] Upgrade async-http-client to 2.12.4 to address CVE-2024-53990 (#23732)
• [fix][sec] Upgrade golang.org/x/crypto from 0.21.0 to 0.31.0 in pulsar-function-go (#23743)
• [improve] Upgrade lombok to 1.18.36 (#23752)
• [improve] Upgrade to Netty 4.1.116.Final and io_uring to 0.0.26.Final (#23813)
• [improve][monitor] Upgrade OTel to 1.45.0 (#23756)

Broker

• [fix][admin] Fix exception loss in getMessageId method (#23766)
• [fix][admin] Fix exception thrown in getMessageId method (#23784)
• [fix][admin] Listen partitioned topic creation event (#23680)
• [fix][admin] Verify is policies read only before revoke permissions on topic (#23730)
• [fix][broker] Add consumer name for subscription stats (#23671)
• [fix][broker] Avoid block markDeletePosition forward when skip lost entries (#21210)
• [fix][broker] Catch exception for entry payload interceptor processor (#23683)
• [fix][broker] Continue using the next provider for authentication if one fails (#23797)
• [fix][broker] Continue using the next provider for http authentication if one fails (#23842)
• [fix][broker] Fix acknowledgeCumulativeAsync block when ackReceipt is enabled (#23841)
• [fix][broker] Fix bug causing loss of migrated information when setting other localPolicies in namespace (#23764)
• [fix][broker] Fix config replicationStartAt does not work when set it to earliest (#23719)
• [fix][broker] Fix deadlock in Key_Shared PIP-379 implementation (#23854)
• [fix][broker] Fix enableReplicatedSubscriptions (#23781)
• [fix][broker] Fix items in dispatcher.recentlyJoinedConsumers are out-of-order, which may cause a delivery stuck (#23802)
• [fix][broker] Fix possible mark delete NPE when batch index ack is enabled (#23833)
• [fix][broker] Fix the retry mechanism in MetadataCache#readModifyUpdateOrCreate (#23686)
• [fix][broker] Invoke custom BrokerInterceptor's onFilter method if it's defined (#23676)
• [fix][broker] Msg delivery is stuck due to items in the collection recentlyJoinedConsumers are out-of-order (#23795)
• [fix][broker] PIP-399: Fix Metric Name for Delayed Queue (#23712)
• [fix][broker] Remove blocking calls from internalGetPartitionedStats (#23832)
• [fix][broker] Remove failed OpAddEntry from pendingAddEntries (#23817)
• [fix][broker] Revert "[fix][broker] Cancel possible pending replay read in cancelPendingRead (#23384)" (#23855)
• [fix][broker] System topic should not be migrated during blue-green cluster migration (#23767)
• [fix][broker] fix NPE when calculating a topic's backlogQuota (#23720)
• [fix][broker] support missing cluster level fine-granted permissions (#23675)
• [fix][broker] topic policy deadlock block metadata thread. (#23786)
• [fix][ml] Topic load timeout due to ml data ledger future never finishes (#23772)
• [improve][admin] Opt-out of topic-existence check (#23709)
• [improve][broker] Improve SystemTopicBasedTopicPoliciesService reader to reduce GC pressure (#23780)
• [improve][broker] Optimize subscription seek (cursor reset) by timestamp (#22792)
• [improve][broker] Reduce unnecessary REPLICATED_SUBSCRIPTION_SNAPSHOT_REQUEST (#23839)
• [improve][log] Print ZK path if write to ZK fails due to data being too large to persist (#23652)
• [improve][ml] Optimize BlobStoreManagedLedgerOffloader.getOffloadPolicies (#23776)
• [improve][txn] Improve Reader in TransactionBuffer to reduce GC pressure (#23779)

Client

• [Fix][Client] Fix pending message not complete when closeAsync (#23761)
• [fix][client] Cannot access message data inside ProducerInterceptor#onSendAcknowledgement (#23791)
• [fix][client] Fix compatibility between kerberos and tls (#23798)
• [fix][client] Fix enableRetry for consumers using legacy topic naming where cluster name is included (#23753)
• [fix][client] Fix memory leak when publishing encountered a corner case error (#23738)
• [fix][client] Fix reader message filtering issue during blue-green cluster switch (#23693)
• [fix][client] Fix wrong start message id when it's a chunked message id (#23713)
• [fix][client] Make DeadLetterPolicy & KeySharedPolicy serializable (#23718)
• [fix][client] Orphan producer when concurrently calling producer closing and reconnection (#23853)
• [fix][client] Prevent retry topic and dead letter topic producer leaks when sending of message fails (#23824)
• [fix][doc] Refine ClientBuilder#memoryLimit and ConsumerBuilder#autoScaledReceiverQueueSizeEnabled javadoc (#23687)
• [improve][client] Make replicateSubscriptionState nullable (#23757)
• [improve][client] PIP-393: Improve performance of Negative Acknowledgement (#23600)
• [improve][client] PIP-393: Support configuring NegativeAckPrecisionBitCnt while building consumer. (#23804)
• [improve][client] Print consumer stats log if prefetched messages are not zero (#23698)

Pulsar IO and Pulsar Functions

• [improve][fn] Improve closing of producers in Pulsar Functions ProducerCache invalidation (#23734)
• [improve][fn] Improve implementation for maxPendingAsyncRequests async concurrency limit when return type is CompletableFuture<Void> (#23708)
• [improve][io] Bump io.lettuce:lettuce-core from 5.0.2.RELEASE to 6.5.1.RELEASE in /pulsar-io/redis (#23685)

Others

• [fix][common] TopicName: Throw IllegalArgumentException if localName is whitespace only (#23691)
• [fix] Fix issues with Pulsar Alpine docker image stability: remove glibc-compat (#23762)
• [fix][cli] Fix set topic retention policy failed (#23688)
• [fix][cli] Fix set-retention with >2GB size value for topic policy (#23689)
• [fix][misc] Honor dynamic log levels in log4j2.yaml (#23847)
• [improve] Sup...

v3.3.4

2025-01-20

Library updates

• [improve] Upgrade lombok to 1.18.36 (#23752)
• [improve] Upgrade to Netty 4.1.116.Final and io_uring to 0.0.26.Final (#23813)
• [improve][monitor] Upgrade OTel to 1.45.0 (#23756)
• [fix][sec] Bump commons-io version to 2.18.0 (#23684)
• [fix][sec] Mitigate CVE-2024-53990 by disabling AsyncHttpClient CookieStore (#23725)
• [fix][sec] Upgrade async-http-client to 2.12.4 to address CVE-2024-53990 (#23732)
• [fix][sec] Upgrade golang.org/x/crypto from 0.21.0 to 0.31.0 in pulsar-function-go (#23743)

Broker

• [feat][broker] Implement allowBrokerOperationAsync in PulsarAuthorizationProvider to avoid exception thrown (#23663)
• [fix][admin] Fix exception loss in getMessageId method (#23766)
• [fix][admin] Fix exception thrown in getMessageId method (#23784)
• [fix][broker] Add consumer name for subscription stats (#23671)
• [fix][broker] Catch exception for entry payload interceptor processor (#23683)
• [fix][broker] Continue using the next provider for authentication if one fails (#23797)
• [fix][broker] Fix bug causing loss of migrated information when setting other localPolicies in namespace (#23764)
• [fix][broker] Fix config replicationStartAt does not work when set it to earliest (#23719)
• [fix][broker] Fix enableReplicatedSubscriptions (#23781)
• [fix][broker] Fix items in dispatcher.recentlyJoinedConsumers are out-of-order, which may cause a delivery stuck (#23802)
• [fix][broker] Invoke custom BrokerInterceptor's onFilter method if it's defined (#23676)
• [fix][broker] Remove failed OpAddEntry from pendingAddEntries (#23817)
• [fix][broker] Skip to persist cursor info if it failed by cursor closed (#23615)
• [fix][broker] System topic should not be migrated during blue-green cluster migration (#23767)
• [fix][broker] fix NPE when calculating a topic's backlogQuota (#23720)
• [fix][broker] support missing cluster level fine-granted permissions (#23675)
• [fix][broker] support missing tenant level fine-granted permissions (#23660)
• [fix][broker] support missing broker level fine-granted permissions (#23637)
• [fix][broker] Msg delivery is stuck due to items in the collection recentlyJoinedConsumers are out-of-order (#23795)
• [fix][common] TopicName: Throw IllegalArgumentException if localName is whitespace only (#23691)
• [fix][ml] Topic load timeout due to ml data ledger future never finishes (#23772)
• [improve][admin] Opt-out of topic-existence check (#23709)
• [improve][broker][branch-3.3] Optimize PersistentTopic.getLastDispatchablePosition (#22707) (#23826)

Client

• [Fix][Client] Fix pending message not complete when closeAsync (#23761)
• [fix][client] Cannot access message data inside ProducerInterceptor#onSendAcknowledgement (#23791)
• [fix][client] Fix enableRetry for consumers using legacy topic naming where cluster name is included (#23753)
• [fix][client] Fix memory leak when publishing encountered a corner case error (#23738)
• [fix][client] Fix reader message filtering issue during blue-green cluster switch (#23693)
• [fix][client] Make DeadLetterPolicy & KeySharedPolicy serializable (#23718)
• [fix][client] Prevent retry topic and dead letter topic producer leaks when sending of message fails (#23824)
• [improve][client] Make replicateSubscriptionState nullable (#23757)

Pulsar IO and Pulsar Functions

• [improve][fn] Improve closing of producers in Pulsar Functions ProducerCache invalidation (#23734)
• [improve][fn] Improve implementation for maxPendingAsyncRequests async concurrency limit when return type is CompletableFuture (#23708)

Others

• [fix][cli] Fix set topic retention policy failed (#23688)
• [fix][cli] Fix set-retention with >2GB size value for topic policy (#23689)
• [fix] Fix issues with Pulsar Alpine docker image stability: remove glibc-compat (#23762)
• [improve] Install coreutils in docker image to improve compatibility (#23667)

Tests & CI

• [fix][test] Remove useless test code (#23823)
• [fix][test]: Flaky-test: GetPartitionMetadataMultiBrokerTest.testCompatibilityDifferentBrokersForNonPersistentTopic (#23666)

For the complete list, check the full changelog.

v3.0.9

2025-01-20

Library updates

• [fix][sec] Bump commons-io version to 2.18.0 (#23684)
• [fix][sec] Mitigate CVE-2024-53990 by disabling AsyncHttpClient CookieStore (#23725)
• [fix][sec] Upgrade async-http-client to 2.12.4 to address CVE-2024-53990 (#23732)
• [fix][sec] Upgrade golang.org/x/crypto from 0.21.0 to 0.31.0 in pulsar-function-go (#23743)
• [improve] Upgrade lombok to 1.18.36 (#23752)
• [improve] Upgrade to Netty 4.1.116.Final and io_uring to 0.0.26.Final (#23813)

Broker

• [fix][broker] Msg delivery is stuck due to items in the collection recentlyJoinedConsumers are out-of-order (#23795)
• [fix][broker] Add consumer name for subscription stats (#23671)
• [fix][broker] Catch exception for entry payload interceptor processor (#23683)
• [fix][broker] Continue using the next provider for authentication if one fails (#23797)
• [fix][broker] Fix config replicationStartAt does not work when set it to earliest (#23719)
• [fix][broker] Fix enableReplicatedSubscriptions (#23781)
• [fix][broker] Fix items in dispatcher.recentlyJoinedConsumers are out-of-order, which may cause a delivery stuck (#23802)
• [fix][broker] Remove failed OpAddEntry from pendingAddEntries (#23817)
• [fix][broker] Skip to persist cursor info if it failed by cursor closed (#23615)
• [fix][broker] fix NPE when calculating a topic's backlogQuota (#23720)
• [fix][admin] Fix exception loss in getMessageId method (#23766)
• [fix][admin] Fix exception thrown in getMessageId method (#23784)
• [fix][admin] Listen partitioned topic creation event (#23680)
• [improve][admin] Opt-out of topic-existence check (#23709)
• [improve][log] Print ZK path if write to ZK fails due to data being too large to persist (#23652)
• [fix][ml] Topic load timeout due to ml data ledger future never finishes (#23772)

Client

• [Fix][Client] Fix pending message not complete when closeAsync (#23761)
• [fix][client] Cannot access message data inside ProducerInterceptor#onSendAcknowledgement (#23791)
• [fix][client] Fix enableRetry for consumers using legacy topic naming where cluster name is included (#23753)
• [fix][client] Fix memory leak when publishing encountered a corner case error (#23738)
• [fix][client] Fix wrong start message id when it's a chunked message id (#23713)
• [fix][client] Make DeadLetterPolicy & KeySharedPolicy serializable (#23718)
• [fix][client] Prevent retry topic and dead letter topic producer leaks when sending of message fails (#23824)
• [fix][client][branch-3.0] Fix compatibility between kerberos and tls (#23801)
• [improve][client] Make replicateSubscriptionState nullable (#23757)
• [fix][doc] Refine ClientBuilder#memoryLimit and ConsumerBuilder#autoScaledReceiverQueueSizeEnabled javadoc (#23687)

Pulsar IO and Pulsar Functions

• [improve][fn] Improve closing of producers in Pulsar Functions ProducerCache invalidation (#23734)
• [improve][fn] Improve implementation for maxPendingAsyncRequests async concurrency limit when return type is CompletableFuture (#23708)
• [improve][io] Bump io.lettuce:lettuce-core from 5.0.2.RELEASE to 6.5.1.RELEASE in /pulsar-io/redis (#23685)
• [fix][fn][branch-3.0] Fix pulsar-function-go compilation

Others

• [fix][common] TopicName: Throw IllegalArgumentException if localName is whitespace only (#23691)

Tests & CI

• [fix][test] Remove useless test code (#23823)
• [fix][test]: Flaky-test: GetPartitionMetadataMultiBrokerTest.testCompatibilityDifferentBrokersForNonPersistentTopic (#23666)

For the complete list, check the full changelog.

v3.3.3

2024-12-03

Library updates

• [fix][sec] Drop hdfs2 support, Upgrade hadoop3 to 3.4.0 and dnsjava to 3.6.2 to address CVE-2024-25638 (#23411)
• [fix][sec] Replace bcprov-jdk15on dependency with bcprov-jdk18-on (#23532)
• [fix][sec] Upgrade Zookeeper to 3.9.3 to address CVE-2024-51504 (#23581)
• [fix][sec] Upgrade to Netty 4.1.115.Final to address CVE-2024-47535 (#23596)
• [improve] Upgrade OpenTelemetry library to 1.44.1 version (#23656)
• [improve] Upgrade oxia-java to 0.4.10 and fix closing of OxiaMetadataStore (#23653)

Broker

• [fix][broker] Avoid orphan ledgers in BucketDelayedDeliveryTracker (#22802)
• [fix][broker] Broker is failing to create non-durable sub if topic is fenced (#23579)
• [fix][broker] Fix AvgShedder strategy check (#23156)
• [fix][broker] Fix Broker migration NPE while broker tls url not configured (#23534)
• [fix][broker] Fix currently client retries until operation timeout if the topic does not exist (#23530)
• [fix][broker] Fix failed TokenAuthenticatedProducerConsumerTest (#23602)
• [fix][broker] Fix namespace unload might be blocked too long with extensible load manager (#23433) (#23517)
• [fix][broker] Fix ownership loss (#23515)
• [fix][broker] Fix print cluster migration state response (#23535)
• [fix][broker] Fix race-condition causing repeated delete topic (#23522)
• [fix][broker] Fix the broker registering might be blocked for long time (#23371) (#23507)
• [fix][broker] Fix unloadNamespaceBundlesGracefully can be stuck with extensible load manager (#23349) (#23496)
• [fix][broker] Increase readBuffer size for bookkeeper.DLOutputStream (#23548)
• [fix][broker] Topics failed to delete after remove cluster from replicated clusters set and caused OOM (#23360)
• [fix][broker] fix logging with correct error message while loading the topic (#23544)
• [fix][broker] fix null lookup result when brokers are starting (#23642)
• [fix][broker] normalize path (#23438)
• [fix][broker] timeout when broker registry hangs and monitor broker registry (ExtensibleLoadManagerImpl only) (#23382) (#23510)
• [improve][broker] Add log to track issue when handleGetTopicsOfNamespace (#23434)
• [improve][broker] Clear thread local BrokerEntryMetadata instance before reuse (#22752)
• [improve][broker] Decouple pulsar_storage_backlog_age_seconds metric with backlogQuota check (#23619)
• [improve][broker] Exclude system topics from namespace level publish and dispatch rate limiting (#23589)
• [improve][broker] Make cluster metadata init command support metadata config path (#23269)
• [improve][broker] Make cluster metadata teardown command support metadata config path (#23520)
• [improve][broker] PIP-383: Support granting/revoking permissions for multiple topics (#23372)
• [improve][broker] PIP-392: Add configuration to enable consistent hashing to select active consumer for partitioned topic (#23584)
• [improve][broker] Skip unloading when bundle throughput is zero (ExtensibleLoadManagerImpl only) (#23626)
• [improve][broker] Support cleanup replication cluster and allowed cluster when cluster metadata teardown (#23561)
• [improve][broker] re-elect the channel owner if no channel owner is found (#23516) (#23580)
• [improve][broker] replace HashMap with inner implementation ConcurrentLongLongPairHashMap in Negative Ack Tracker. (#23582)

Client

• [fix][admin] Fix lookup get a null result if uses proxy (#23556)
• [fix][client] Fix DLQ producer name conflicts when there are same name consumers (#23577)
• [fix][client] Fix Reader.hasMessageAvailable return wrong value after seeking by timestamp with startMessageIdInclusive (#23502)
• [fix][client] Fix ReaderBuilder doest not give illegalArgument on connection failure retry (#22639)
• [fix][client] Fix deadlock of NegativeAcksTracker (#23651)
• [fix][client] Fix producer/consumer stop to reconnect or Pub/Sub due to IO thread race-condition (#23499)
• [fix][client] Fix race-condition causing doReconsumeLater to hang when creating retryLetterProducer has failed (#23560)
• [fix][client] Fix the javadoc for ConsumerBuilder.isAckReceiptEnabled (#23452)
• [fix][client] Fixed an issue where a cert chain could not be used in TLS authentication (#23644)
• [fix][client] Initializing client-authentication using configured auth params (#23610)
• [fix][client] Make protobuf-java dependency optional in java client libraries (#23632)
• [fix][client] Prevent embedding protobuf-java class files in pulsar-client-admin and pulsar-client-all (#23468)
• [fix][client] The partitionedProducer maxPendingMessages always is 0 (#23593)
• [fix][client] Use dedicated executor for requests in BinaryProtoLookupService (#23378)
• [fix][client] fix incomingMessageSize and client memory usage is negative (#23624)
• [fix][client] fix the beforeConsume() method earlier hit with message listener (#23578)
• [improve][admin] Print error log if handle http response fails (#23563)
• [improve][client] Enhance error handling for non-exist subscription in consumer creation (#23254)
• [improve][client] Increase default Java client connectionMaxIdleSeconds to 60 seconds (#23430)
• [improve][client] Reduce unshaded dependencies and shading warnings in shaded Java client modules (#23647)
• [improve][client] Replace NameUtil#generateRandomName with RandomStringUtils#randomAlphanumeric (#23645)

Pulsar IO and Pulsar Functions

• [fix][fn] ack messages for window function when its result is null (#23618)
• [improve][io] Support update subscription position for sink connector (#23538)
• [improve][io] Upgrade Spring version to 6.1.13 in IO Connectors (#23459)

Others

• [fix][misc] Class conflict during jetcd-core-shaded shading process (#23641)
• [fix][misc] Unable to connect an etcd metastore with recent releases due to jetc-core sharding problem (#23604)
• [fix][ml] Managed ledger should recover after open ledger failed (#23368)
• [fix][proxy] Fix pattern consumer does not work when using Proxy (#23489)
• [fix][ws] Implement missing http header data functions in AuthenticationDataSubscription (#23638)
• [improve] Improve logic for enabling Netty leak detection (#23613)
• [improve] Use single buffer for metrics when noUnsafe use (#23612)
• [improve][misc] Disable OTel by default when running the pulsar-perf tool (#23585)
• [improve][misc] Upgrade Jetty to 9.4.56.v20240826 (#23405)
• [improve][ml] Avoid repetitive nested lock for isMessageDeleted in ManagedCursorImpl ([#23609](htt...