Cause analysis: starting yarn-session.sh will initiate a request to port 8032 of resourcemanager:
However, port 8032 has been unable to be requested, and the retry mechanism will be triggered and will continue to try.
Remarks: When this problem occurred, the HA high-availability hadoop deployed by ambari in my environment had three nodes node104, node105, and node106. Among them, node105 and node106 are resourcemanager carriers, node106 is active and node105 is in standby.
Possible causes of this problem:
1. The hosts configuration file on each node may be incorrect or the hostname may be inconsistent.
2. Firewall reasons
3. The environment variables on the node are configured incorrectly. My environment variable configuration file is /etc/profile.
My reason is: export HADOOP_CLASSPATH=`hadoop classpath` in the environment variable configuration file /etc/profile was mistakenly written as export HADOOP_CLASSPATH='hadoop classpath'
After the correction, yarn-session was successfully started.