All of them are data nodes 2 are name nodes with high availability setup .
When restarting the name node service from cloudera manager the name node switch to the standby node with no issue .
But when shuting down the name node machine .I woul expect that the cluster will continue to function but is seems that the commands that use the name service fail as it search for the the sacond name node machine .
This is the error I get when do ls on hdfs dir :
root@gc-test-impala28-03:~# hadoop fs -ls /user 17/06/20 13:33:32 WARN hdfs.DFSUtil: Namenode for nameservice1 remains unresolved for ID namenode566. Check your hdfs-site.xml file to ensure namenodes are configured properly. -ls: java.net.UnknownHostException: gc-test-impala28-04.c.bi-environment1271.internal Usage: hadoop fs [generic options] -ls [-C] [-d] [-h] [-q] [-R] [-t] [-S] [-r] [-u] [<path> ...]
gc-test-impala28-04 is the stopped name node .
Does HDFS high availability really suppert machine failure or only failed service ?
The cluster have :
3 zookeeper instances (One on the failed namenode)