We just ran into this problem. @Jeff Arnold above is correct that since the standby namenode is down the dfsadmin commands will fail. Instead of the doing the /etc/hosts file change he recommends you can manually override the -fs in the commands suggested here https://issues-test.apache.org/jira/browse/HDFS-8277?focusedCommentId=14517247&page=com.atlassian.jira.plugin.system.issuetabpanels%3Acomment-tabpanel#comment-14517247. The dfsadmin commands change to this for example sudo -u hdfs hdfs dfsadmin -fs hdfs://<active_namenode>:<rpc_port> -safemode enter Also if you are using Cloudera Manager the config that gets used by "namenode -bootstrapStandby" command does not include the necessary config for the journal nodes for shared edits. You will need to copy the running config from the running active namenode. It will be under something like /run/cloudera-scm-agent/process/5134-hdfs-NAMENODE. Copy that to the standby namenode and set the bootstrap command to use it. sudo -i -u hdfs HADOOP_CONF_DIR=<your_copied_config> hdfs namenode -bootstrapStandby -force ... View more