v8.5.18

Compare Source

version:8.5.18

Bug Fixes

Misc

• Exception message on unexpected instance migration error displays the wrong key (#​31541)
• Warnings related to bean is not eligible for getting processed by all BeanPostProcessors and MeterFilter is being configured after a Meter has been registered to this registry. (#​30812)
• Do not use KeyGenerator in scheduled task (#​30536)
• Inter-partition message correlation failing when the partition count is changed in the configuration (#​29612)
• Output mappings applied before updating multi instance body output collection (#​23658)

Merged Pull Requests

• [stable/8.6] fix: removal of meter registry workaround (#​31168)
• [Backport stable/8.7] fix: when there is a leader transition client requests were never completed (#​31141)
• Backport to stable/8.5: do not reconfigure default ObjectMapper instance (#​31071)
• fix: Quote changelog variable (#​28426) (#​31043)
• fix: when there is a leader transition client requests were never completed (#​31033)
• test: assert that correlationKey is not associated to partition 1 (#​30967)
• fix: Use vault-based token in Maven Release job (#​28522) (#​30605)
• fix: StagedCache should only cache configured intents (#​30560)
• deps: upgrade springboot to 3.4.4 for stable/8.6 (#​30533)
• feat: initialize BrokerInfo from config so that they are used before bootstrapping (#​30377)
• [Backport stable/8.7] fix: do not initialize BrokerInfo from config (#​30337)
• ci: new github workflow for license analysis with FOSSA (#​30246)
• fix: do not initialize BrokerInfo from config (#​30217)
• fix: persisted partition count is used in EngineProcessors (#​30110)
• fix: (8.6) bean post processor produced warn logs (#​28547)
• Output mappings applied before updating multi instance body output collection (#​25106)

v8.5.17

Compare Source

version:8.5.17

Bug Fixes

Misc

• Pollution of command cache breaks scheduled tasks (#​29735)
• Possible to deploy process for the default tenant without access (#​29236)
• Optimize UpdateIndexStepResumesReindexOperationsIT is failing for OpenSearch (#​29004)
• Banning escape hatch doesn't work as expected on termination for recursive process models (#​16410)
• Slow processing of large multi instance collection (#​12946)

Merged Pull Requests

• Small optimization in BufferUtil: do not allocate an empty string if buffer is empty (#​30460)
• Revert "ci: specify permissions for release workflow" (#​30420)
• ci: specify permissions for release workflow (#​30411)
• fix: do not throw a NullPointerException when the firstSegment is null in SegmentedJournal (#​30234)
• refactor: add git commit info to built artifact (#​29870)
• fix: don't pollute the command cache (#​29769)
• feature: Increase the default inbound message size to 5MB for clients (#​29608)
• Fix process instance batch events and convert to looping mechanism (#​29310)
• Require access to default tenant to use deprecated DeployProcess RPC (#​29239)
• Create meters once to avoid gauge re-registration (#​29136)
• Avoid warning when closing OS exporter if open was not called (#​29133)
• fix: don't try to write responses without request id (#​29002)
• ci: remove broken benchmark measurement support (#​27911)
• Reject pull request updates using merge commits (#​23274)
• Raise incident at specific called process depth (#​22735)

v8.5.16

Compare Source

version:8.5.16

Bug Fixes

Misc

• NettyDnsMetrics is not safe under concurrent access (#​28556)
• Message correlation on a non-interrupting message event is not idempotent (#​26926)

Maintenance

• Auto-assign API reviewers to OpenAPI spec changes (#​28384)

Merged Pull Requests

• [Backport stable/8.6] deps: Update dependency org.apache.httpcomponents.client5:httpclient5 to v5.4.1 (#​28724)
• [Backport stable/8.6] Update dependency org.apache.httpcomponents.core5:httpcore5 to v5.3 (#​28723)
• Assign original author to backport pull requests (#​28671)
• fix: NettyDnsMetrics use a MeterProvider instead of a HashMap to avoid concurrency issues (#​28555)
• Fix NPE when access a null code and reusing the same metric name for different metrics (#​28480)
• Create CODEOWNERS for OpenAPI changes (#​28385)
• test: verify that passive director closes all exporters (#​28310)
• fix: always close exporter container and distribution service (#​28243)
• deps: Update dependency org.apache.httpcomponents.client5:httpclient5 to v5.4.1 (#​27807)
• [Backport stable/8.7] deps: Update dependency org.apache.httpcomponents.core5:httpcore5 to v5.3 (main) (#​27764)
• Idempotent message correlation retries (#​27525)

v8.5.15

Compare Source

Bug Fixes

Misc

• Non-Prometheus Micrometer implementations only expose few C8 metrics (#​26078)

Merged Pull Requests

• fix: RaftRequestMetrics is threadsafe (#​28057)
• migrate TopologyMetrics to micrometer (#​27986)
• Migrate zeebe-backup to micrometer (#​27907)
• refactor: use set for clarity (#​27876)
• migrate MessagingMetrics to micrometer (#​27848)
• migrate RaftServiceMetrics & RaftStartupMetrics to micrometer (#​27830)

v8.5.14

Compare Source

version:

Bug Fixes

Misc

• Non-Prometheus Micrometer implementations only expose few C8 metrics (#​26078)
• UnsupportedOperationException: Incorrect usage of actor.call(...) cannot be called from current actor. (#​25897)

Maintenance

• [FEEL] Store duration values in a normalized format (#​27098)
• Replace @​AutoCloseResources extension with new built-in Junit AutoClose extension (#​26421)

Merged Pull Requests

• migrate JournalMetrics to micrometer (#​27929)
• Ignore misconfigured execution listeners (#​27904)
• Migrate gateway/gRPC metrics to Micrometer (#​27878)
• Fix setting bounded cache size metric (#​27875)
• migrate RaftServiceMetrics & RaftStartupMetrics to micrometer (#​27830)
• Migrate process engine metrics to Micrometer (#​27790)
• migrate RaftRequestMetrics to micrometer (#​27769)
• Introduce a Map with 3 keys (Map3D) as a generic collection (#​27759)
• migrate MetaStoreMetrics/SnapshotReplicationMetrics to prometheus (#​27754)
• Migrate RaftReplicationMetrics to micrometer (#​27732)
• Migrate RaftRoleMetrics to micrometer (#​27731)
• migrate LeaderMetrics to micrometer (#​27728)
• Pass and store meter registry down to ProcessEngineMetrics (#​27686)
• feat: allocate a partition MeterRegistry in RaftPartition (#​27684)
• Migrate stream processor metrics to Micrometer (#​27669)
• Migrate ProcessingMetrics from stream platform to Micrometer (#​27654)
• Migrate replay metrics of stream platform to micrometer (#​27650)
• Migrate partition role metrics to Micrometer (#​27644)
• fix: allow unauthenticated api access by default (#​27608)
• Migrate stream processor metrics to Micrometer (#​27564)
• Migrate replay metrics of stream platform to micrometer (#​27561)
• Migrate broker health metrics to Micrometer (#​27538)
• Migrate bounded command cache metrics to Micrometer (#​27536)
• Migrate job stream metrics to Micrometer (#​27534)
• Migrate exporter director metrics to Micrometer (#​27532)
• fix: Store duration values in a normalized format (#​27161)
• fix: avoid actor call to unregister handlers when it is closing (#​26759)
• deps: Update dependency software.amazon.awssdk:bom to v2.29.48 (main) (#​26708)
• Resolve junit 5 maven dependency convergence (#​26641)

v8.5.13

Compare Source

version:8.5.13

Dependency Updates

• deps: Update eclipse-temurin:21-jdk-noble Docker digest to 971bf64 (stable/8.5) by @​renovate in https://github.com/camunda/camunda/pull/27421
• deps: Update dependency io.swagger.core.v3:swagger-annotations to v2.2.28 (stable/8.5) by @​renovate in https://github.com/camunda/camunda/pull/27039
• deps: Update dependency io.camunda:zeebe-qa to v8.5.12 (stable/8.5) by @​renovate in https://github.com/camunda/camunda/pull/27052
• deps: Update dependency com.azure:azure-sdk-bom to v1.2.31 (stable/8.5) by @​renovate in https://github.com/camunda/camunda/pull/27494
• deps: Update dependency com.google.protobuf:protobuf-bom to v3.25.6 (stable/8.5) by @​renovate in https://github.com/camunda/camunda/pull/27498

v8.5.12

Compare Source

version:8.5.12

Fixes

• fix: @​TestZeebe resources are closed even when constructed in a @​BeforeEach (#​26737)
• Fixes Zeebe 8.5 Limiters inFlight concurrent access (#​26408)

Improvements

• Replace usage of internal Junit 5 helper removed in Junit 5.11 (#​26722)
• ci: use correct latest benchmark version (#​26674)
• ci: update zeebe-performance-test link (#​26128)

v8.5.11

Compare Source

Bug Fixes

Java Client

• Unauthorized REST requests are retried forever when OAuth is used (#​26012)
• Zeebe Java Client Fails to Handle Error Responses Gracefully (#​25087)

Misc

• Same-name form resources inside deployments overridden with unexpected content on deployment distribution (#​25727)
• Channel closed when a response times out in NettyMessagingService (#​25596)
• Process modification on inclusive gateway makes active token disappear (#​25099)
• A process instance is stuck after modifing to an inclusive gateway (#​20587)
• InvalidConfigurationException when GCS store is not accessible on startup (#​14593)

Merged Pull Requests

• test: use lowercase GCS bucketname (#​26538)
• Limit REST requests (#​26442)
• Update zcl repo location (#​26298)
• Support activating Inclusive Gateway via Process Instance Modification (#​25771)
• fix: only create one CREATED event per form on distribution (#​25728)
• fix: parse unexpected REST responses in client (#​25681)
• fix: do not close active channels on first response timeout (#​25615)
• fix: do not throw Exception when GCS store is not accessible on startup (#​23057)

v8.5.10

Compare Source

8.5.10

Bug Fixes

Broker

• Zeebe user tasks not terminated on process instance modification (#​24672)

Java Client

• OAuthCredentialsCache in Zeebe Java client does not work correctly in multi-tenancy setup (#​20471)

Misc

• Form cache is corrupt and causes inconsistent processing (#​25504)
• Process instance no longer subscribed to message after unsuccessful process instance migration (#​25166)
• Retry complete follower append on flush error (#​14867)

Maintenance

• Do not log MessagingException$ConnectionClosed as an error in the gateway (#​18339)

Merged Pull Requests

• fix: don't use mutable keys for forms cache (#​25523)
• [Backport main] Update transient message state as side-effect (#​25476)
• Update transient message state as side-effect (#​25298)
• All Zeebe CI jobs need to specify GHA timeouts (#​24880)
• fix: throw correct error instead of InternalError (#​24680)
• fix: cancel user tasks on instance modification (#​24673)
• fix: switch client credential cache key to clientId (#​24519)
• feature: when a follower fails to flush respond with error to leader (#​24378)
• ci: use non-preemptible runners for jobs with highest disconnect rate (#​24370)
• Properly handle connection closed error in the gRPC server (#​20066)

v8.5.9

Compare Source

Enhancements

Misc

• Allow to cancel bannend instances (#​12772)

Bug Fixes

Misc

• Version update is not persisted in state because transaction is not committed (#​24352)
• Completed zeebe user tasks have string variables without double quotes (#​24217)
• Exported record with duplicate sequence number (#​24192)
• Number of threads created in Gateway for blocking identity calls causing "too many concurrent streams" error (#​23853)
• Partition backup fails because of concurrent log compaction (#​23082)
• StackOverflow during process cancelation (#​21991)
• fix/Spring Boot SDK: do not ignore json field names from annotations (#​21962)
• JobRecord elementId gets overwritten when an error is thrown during job execution (#​21567)
• Timers of banned instances are forever re-triggered (#​14213)
• Error on retrieving write buffer from log stream (#​10141)

Maintenance

• Backport {operate,optimize, identity}/client/.gitignore into stable/8.5, stable/8.4, stable/8.3 branches (#​24311)
• Improve error mapping of REST API (#​17441)

Merged Pull Requests

• fix: commit last transaction after running migrations (#​24353)
• Prevent duplicate sequence numbers in records exported to Elasticsearch and OpenSearch (#​24214)
• fix: set the correct elementId in Incident record when Job throw with no catch event (#​24206)
• Fix identity-sdk concurrency issue (#​24196)
• feat: backup only segment containing data not present in snapshots (#​24184)
• fix: enable configuration processor (#​24153)
• Removing recursion from process cancellation logic (#​24082)
• Add actor clock endpoint to test framework (#​24018)
• fix: compare with base branch or previous commit (#​23860)
• Allow cancelling banned instances (#​23849)
• fix: include stack trace on failed GRPC token verification (#​23600)
• fix: complete future with error when leadership change is cancelled (#​23371)
• Use overrides from @​JsonProperty when it's provided (#​23219)
• Improve REST API error mapping (#​22412)
• fix: merge concurrent resolves for up to 128 hostnames (#​22410)