Improve rebalance phase time spent during the conflict resolution ----------------------------------------------------------------- Key: ISPN-11099 URL: https://issues.redhat.com/browse/ISPN-11099 Project: Infinispan Issue Type: Enhancement Components: Core Affects Versions: 10.1.0.CR1 Reporter: Diego Lovison Priority: Critical Attachments: log.txt Give a sync replica cache with 3 nodes Introduce a network partition Change the cache value Fix the network partition Check the new cache value https://github.com/diegolovison/jgroups-chaos mvn test -Dtest=InfinispanPartitionHandlingTest -Dlog4j.configurationFile=log4j2-ispn-conflict-manager.xml config: {code:java} cacheConfigurationBuilder.clustering().cacheMode(CacheMode.REPL_SYNC) .partitionHandling().mergePolicy(MergePolicy.REMOVE_ALL); {code} logs: {noformat} 15:44:31,827 INFO (stateTransferExecutor-thread--p22-t1) [CLUSTER] [Context=org.infinispan.CONFIG] ISPN100002: Starting rebalance with members [Diegos-MacBook-Pro-40446, Diegos-MacBook-Pro-6797, Diegos-MacBook-Pro-52614], phase READ_OLD_WRITE_ALL, topology id 14 15:44:56,725 TRACE (stateTransferExecutor-thread--p22-t1) [StateReceiverImpl] Cache fooCache attempting to receive replicas for segment 83 from [Diegos-MacBook-Pro-40446, Diegos-MacBook-Pro-6797, Diegos-MacBook-Pro-52614] with topologyId=12, timeout=215112 {noformat} expected result: for one key, running all cluster in the same machine. I think that we can improve the business rules to avoid taking more than 10 seconds to finish the test.