Coherence CE v23.03
This is the official Coherence CE v23.03 release.
Requirements:
- Java 17 or later
New Features in CE 23.03:
- CDI Response Caching - Use caching annotations for response caching in CDI enabled applications such as Helidon
- Topics Management Improvements - This version of Coherence introduces additional features and functionality to help you manage and monitor topics within a Coherence cluster. Improvements include:
- New MBeans covering Topics, Subscribers and Subscriber Groups
- Reports for the above MBeans
- Additional Management over REST endpoints for Topics, Subscribers and Subscriber Groups
- Microprofile Health - Coherence MP Health is a very simple module that enables you to publish Coherence health checks into the MicroProfile Health Check Registries available at runtime
- Gradle POF Plugin - This plugin provides automated instrumentation of classes with the @PortableType annotation to generate consistent (and correct) implementations of Evolvable POF serialization methods
For more information see the Core Improvements documentation.
Bugs fixed since Coherence CE 22.09:
COH-27383Bumped integrated version ofASMto 9.5.COH-27374Fixed an issue that prevented the serialization/deserialization of lambdas with JDK 21.COH-26028Added Topics MBeans.COH-25991Added a new Gradle plugin for POF serialization that instruments classes at build time (similar to the Coherence POF Maven plugin).COH-22849Added CDI support for response caching.COH-27315Fixed an issue where callingclear()on a cache from a gRPC client removed entries using a synthetic delete instead of a real delete, and hence appears as an eviction.COH-27249Fixed poor performance inSafeSortedMapmethodsgetEntrySet/getEntrywhen there are a large number of entries in the map and ensured noSafeSortedMapmethod returnsSafeSortedMap.NULLfor an entry key or value.COH-27205Fixed an issue where Coherence fails to load a cache configuration file if the<reconnect-interval>for a<view-scheme>was in form of a time unit, e.g. "30s".COH-27186Fixed an issue wherePortableTypeGeneratormay generate an incorrect implementation for "public EvolvablegetEvolvable(int nTypeId)".COH-26665Improved parallelism of queries, aggregations and bulk entry processor requests by splitting them by partition, instead of by member.COH-27094Fixed an issue where when the management senior leaves the cluster registered health checks could disappear on the new management senior.COH-27086Fixed an issue where aPartitionedCacheservice may terminate unexpectedly due to an unhandledArrayIndexOutOfBoundsExceptionbeing thrown while processing anUpdateIndexRequest.COH-27085Fixed a potential thread deadlock where an initializingContinuousQueryCachereceives and attempts to process a cache truncation event.COH-27046Fixed an issue where some reporter group files do not honor thecoherence.reporter.frequencysystem property.COH-27001Fixed a race condition in the concurrent executor serviceRecordingSubscriberwhich can result in theRecordingSubscriberreporting an incorrect state.COH-26997Added integration with the Microprofile Health API so server Coherence health checks via MP Health endpoints.COH-26992Added system propertiescoherence.publisher.resend.intervalandcoherence.publisher.delivery.timeoutfor setting the packet delivery<resend-milliseconds>and<timeout-milliseconds>element values.COH-26958Includedcom.tangosol.net.sslandcom.oracle.coherence.caffeinepackages in JavaDoc generation.16.COH-26936Added a new report capturing usage statistics forTopicsCOH-26919Fixed an issue where theServiceMBean.TaskMaxBacklogmight not be updated to have the maximum task backlog.COH-26903Changed the<shutdown-listener>default fromforcetogracefuland added system propertycoherence.shutdown.timeoutwith a default of 2 minutes.COH-26868Fixed an issue where multiple topic subscribes joining and leaving simultaneously could cause Coherence key contention deadlock detection to be triggered.COH-26764Fixed an issue with where cancelling aCompletableFuturereturned by aNamedTopicSubscriber stopped the subscriber from receiving more messages. Cancelled or completed futures are now handled correctly.COH-26758Bumped Helidon (and associated dependencies) to 3.0.2.COH-26747Updated Javadoc to reflect the classes that had moved fromcom.oracle.commontocom.oracle.coherence.common.COH-26739Fixed an issue wherecoherence.modewould be effectively ignored at the cluster level.COH-26700Fixed an issue where it was not possible to specifically set the serializer for the Coherence system config separately from the defaultcoherence.serializerproperty.COH-26651Fixed an issue in topics where simultaneous joining and/or leaving of subscribers can cause temporary inconsistencies in channel allocations and hence missed messages.COH-26638Fixed an issue where aNullPointerExceptionwould be raised when attempting to get a session name from a remote gRPC Session.COH-26623Added a new report (report-service-storage.xml) which provides partition storage related information.COH-26572Improved JSON serialization ofBig{Decimal,Integer}so these types may be better handled by gRPC clients.COH-26564Implemented periodic flushing of pending messages to reduce native memory usage when sending multiple messages at once.COH-26546Fixed a potentialNullPointerExceptionwhen registering a topic MBeanCOH-26517Updatedprotobuf.versionto 3.21.7.COH-26507Fixed an issue that preventedBigDecimalaggregators to fail deserializing from JSON.COH-26378Fixed an issue where gRPC client connections did not fail over correctly during a rolling restart of the gRPC proxy members in the cluster.COH-26370Fixed an issue where aNullPointerExceptionmay be thrown when aReplicatedCacheservice is shutting down.COH-26355Fixed an issue where the Cache Units attribute or metric could be negative for large caches when the unit factor is greater than 1.COH-26316Changed theNamedCache.entrySet(Filter)implementation to execute query by partition instead of by member, in order to improve parallelism and avoid exceeding the 2GB message limit when executing large queries.COH-26314Enhanced topics to allow the channel count to be changed for an existing topic.COH-26271Improved the cluster member join algorithm to avoid members blocking each other when there are large number of members joining the cluster simultaneously and system resources are under heavy load.COH-26251Fixed an issue where theConnectionManagerMBeanresetStatisticsoperation was missing from REST API and Swagger documentation.COH-26224Fixed an issue where aNearCacheusing invalidation strategypresentfailed to release a key lock withinget/getAll, resulting in a "Detected state corruption on KEY..." log message.COH-26223ImprovedClusterNodeMBean.setLoggingLevel()to apply the change to all supported destinations exceptSLF4Jbecause it does not support the feature.COH-26089Fixed an issue where data loss might occur after multiple rolling restarts when backup persistence is enabled.COH-25984Exposed Topics MBeans via Management Over REST.COH-25823Fixed an issue where theMetricsHttpProxyservice fails to restart on shutdown or unexpected restart of a Coherence member.COH-25604Improved error reporting during the cluster service halting process to help identify what caused the cluster service to halt and any issues that may have been encountered while halting.COH-24792Removed the deprecated classOldCache.COH-24530Fixed an issue whereConnector$Registermay throw aNullPointerExceptionduring a rolling upgrade.COH-23896Fixed an issue whereSafeSortedMapconcurrent access would result in inconsistencies under high stress situations.COH-22165Fixed an issue that could yield ajava.io.StreamCorruptedException: invalid type: 64during a rolling upgrade.