[ https://issues.apache.org/jira/browse/HBASE-10791?page=com.atlassian.jira.plugin.system.issuetabpanels:all-tabpanel ]
Nick Dimiduk updated HBASE-10791: --------------------------------- Attachment: IntegrationTestRegionReplicaPerf.out HBASE-10791.03.patch Patch v3, this one works as described. The following local cluster settings were necessary {noformat} <property> <name>hbase.it.clustermanager.ssh.opts</name> <value>-o StrictHostKeyChecking=no</value> </property> <property> <name>hbase.it.clustermanager.ssh.cmd</name> <value>/usr/bin/ssh %1$s %2$s%3$s%4$s "sudo su - hbase -c \"%5$s\""</value> </property> <property> <name>hbase.master.loadbalancer.class</name> <value>org.apache.hadoop.hbase.master.balancer.StochasticLoadBalancer</value> </property> <property> <name>hbase.regionserver.storefile.refresh.period</name> <value>30000</value> <!-- default hbase.master.cleaner.interval=60000 --> </property> {noformat} For my next trick, I'll work out why the final assertion proves the assumption false. > Add integration test to demonstrate performance improvement > ----------------------------------------------------------- > > Key: HBASE-10791 > URL: https://issues.apache.org/jira/browse/HBASE-10791 > Project: HBase > Issue Type: Sub-task > Components: Performance, test > Affects Versions: hbase-10070 > Reporter: Nick Dimiduk > Assignee: Nick Dimiduk > Attachments: HBASE-10791.00.patch, HBASE-10791.01.patch, > HBASE-10791.02.patch, HBASE-10791.03.patch, > IntegrationTestRegionReplicaPerf.out > > > It would be good to demonstrate that use of region replicas reduces read > latency. PerformanceEvaluation can be used manually for this purpose, but > it's not able to use ChaosMonkey. An integration test can set up the monkey > actions and automate execution. -- This message was sent by Atlassian JIRA (v6.2#6252)