Geode dev team,
I have a total *280 Million rows* in total at source (partitioned) and ingesting in parallel (50+ parallel geode CacheClients doing putALL), Post ingesting half of rows, *one of the servers goes down (crash).* *Additional details* #Servers: 3 (Each having heap approx. 90 GB) //Region creation *create region --name=REGION_NAME --total-num-buckets=283 --type=PARTITION_OVERFLOW --enable-synchronous-disk=false --eviction-action=overflow-to-disk --redundant-copies=0* *What is the default value for eviction-heap-percentage and critical-heap-percentage..? Do I need to explicitly specify while the server starts..?* *Seeking some recommendation to resolve the server down issue, something like,* 1. speed-up eviction (overflow to disk operation) and keep the required heap (run-time memory) available for current putAll operations. 2. reduce GC time (observed large GC time when system/heap is ~50%+ loaded) *GC and gemfire config using while server-startup* GC_SWITCHES_SERVER=" -Xms90000m -Xmx90000m -Xss384K -XX:MaxMetaspaceSize=384m -XX:ParallelGCThreads=4 -XX:NewRatio=2 -XX:SurvivorRatio=8 -XX:CompileThreshold=5000 -XX:+UseConcMarkSweepGC -XX:+UseParNewGC -XX:CMSInitiatingOccupancyFraction=55 -XX:+UseCMSInitiatingOccupancyOnly -XX:+PrintGC -XX:+PrintGCDetails -XX:+PrintGCDateStamps -XX:+PrintGCTimeStamps -XX:+PrintTenuringDistribution -XX:+PrintGCApplicationStoppedTime -XX:+PrintGCApplicationConcurrentTime -Xloggc:/var/geode/cacheServer_gc.log -XX:+UseGCLogFileRotation -XX:NumberOfGCLogFiles=10 -XX:GCLogFileSize=100M -XX:+UnlockDiagnosticVMOptions -XX:ParGCCardsPerStrideChunk=32768" GF_SWITCHES_SERVER=" -Dgemfire.enable-time-statistics=false -Dgemfire.statistic-sampling-enabled=true -DJGroups.SHUN_SUNSET=1 -DBridgeServer.HANDSHAKE_POOL_SIZE=20" Some guidance around required configuration will help. Thanks Nilkanth.
