Created attachment 1073202 [details] Used test code Description of problem: If method RemoteCache.getAll() or RemoteCacheManager.Start() isn't called, server statistics are returning the stats of the same node. Test code in attachament. Version-Release number of selected component (if applicable): 7.2.3.Final of Infinispan packages How reproducible: Always Steps to Reproduce: 1. Start JDG server in clustered mode 2. Wait a while and Start another JDG server in clustered mode 3. Run tests Actual results: Call one: ------------- Server statistics of Hot Rod Server: --------------------- Host: localhost Port: 11222 {currentNumberOfEntries=1000, hits=0, removeHits=0, totalBytesRead=12136, timeSinceStart=15, removeMisses=0, totalNumberOfEntries=489, stores=489, misses=0, retrievals=0, totalBytesWritten=2966} Server statistics of Hot Rod Server: --------------------- Host: localhost Port: 11322 {currentNumberOfEntries=1000, hits=0, removeHits=0, totalBytesRead=12145, timeSinceStart=15, removeMisses=0, totalNumberOfEntries=489, stores=489, misses=0, retrievals=0, totalBytesWritten=3156} Call two: ------------- Server statistics of Hot Rod Server: --------------------- Host: localhost Port: 11222 {currentNumberOfEntries=1000, hits=0, removeHits=0, totalBytesRead=12154, timeSinceStart=16, removeMisses=0, totalNumberOfEntries=489, stores=489, misses=0, retrievals=0, totalBytesWritten=3346} Server statistics of Hot Rod Server: --------------------- Host: localhost Port: 11322 {currentNumberOfEntries=1000, hits=2000, removeHits=0, totalBytesRead=19596, timeSinceStart=12, removeMisses=0, totalNumberOfEntries=511, stores=511, misses=0, retrievals=2000, totalBytesWritten=17008} Tests run: 1, Failures: 0, Errors: 0, Skipped: 0, Time elapsed: 2.893 sec Expected results: That call one will look almost the same as call two. Additional info: Used client server DataGrid with default configuration.
Servers where started from teminal as: Server 1: ./bin/clustered.sh -Djboss.node.name=jdg-0 -Djboss.socket.binding.port-offset=0 Server 2: ./bin/clustered.sh -Djboss.node.name=jdg-1 -Djboss.socket.binding.port-offset=100
We need to add the clustered statistics to the returned stats map and also add a documentation note that the "old" statistics names only refer to the node which handled the request and therefore should be ignored.