Hi Brian,
When the cluster was deployed (i.e. host components got added) the hostname used should be the same that the agents always need to register with. Say "old.host.com" is what the host registered earlier and later it switches to "new.host.com" then Ambari interprets it as: * old.host.com is unavailable * new.host.com is a new host that has joined What you need to do is to make the hosts register with the original name. Afterwards you can DELETE the host entries with the new name. (curl -u admin:admin -H "X-Requested-By: ambari" -i -X DELETE http://AMBARI_SERVER_HOST:8080/api/v1/clusters/c1/hosts/n<http://AMBARI_SERVER_HOST:8080/api/v1/clusters/c1/hosts/HOSTNAME>ew.host.name) If the above is not doable for some reason, reply back on the thread and there are a few more options that we can consider. -Sumit ________________________________ From: Brian Jeltema <[email protected]> Sent: Friday, December 18, 2015 5:15 AM To: [email protected] Cc: Tommy McNeese; Ben Gunter Subject: duplicate host cleanup I just upgraded to Ambari 2.0 on a cluster and ran into a problem with duplicate host definitions. Somehow, some of the cluster nodes were registered twice with different FQDNs: node1.mycompany.com<http://node1.mycompany.com/> node1.foo.mycompany.com<http://node1.foo.mycompany.com/> The correct name is the first one, but if I look at that host in the Ambari Host page, the summary section is showing the second one. It appears this is causing the Ambari server to fail to detected the heartbeat. How can I clean this up? Brian
