This post is regarding a Node installation failed due to a locked Kerberos account. Below the procedure on how to solve this issue.
Symptom
The cluster installation failed during the node installation.
Cause analysis
The /var/log/Bigdata/nodeagent/agentlog/agent.log file contains the following error information.
A node which is not installed this time was previously installed in the cluster and not uninstalled. The node continuously tries to connect to the controller for authentication. As a result, the Kerberos account is locked and other nodes fail to be installed.
View the /var/log/Bigdata/okerberos/oms-krb5kdc.log file on the primary Operation & Maintenance System (OMS) node to check which node continuously tries to connect to the controller for authentication.
In the preceding figure, 10.67.149.73 and 10.67.149.75 are the nodes that were previously installed in the cluster.
Solution
Log in to the node that is not uninstalled, run sh /opt/huawei/Bigdata/nodeagent/setup/ uninstall.sh to uninstall it and reinstall the cluster.
Tips:
In V100R002C80 and later, run sh /opt/huawei/Bigdata/om-agent/nodeagent/setup/uninstall.sh instead.
