Remove Defective Node from a Cloudbreak HDP Cluster

[#1]

Hi all,

I'm running a cloudbreak 1.2 Ambari Cluster whose underlying AWS instances has gone bad. What is the proper way to remove the instance and replace it in the cluster through cloudbreak? Any help or pointing to the relevant documentation would be appreciated.

[#2 · Accepted solution]

Hi @Kenneth Graves,

Please check that the defective node is in a slave/worker hostgroup, because the following solution based on Cloudbreak upscale functionality that only works on slave/worker hostgroups:

First you should sync the cluster state, sync could be triggered through the Cloudbreak UI when the "cluster details" panel has been opened.

After the sync has been applied the defective node should be marked on the node list of the "cluster details" panel and a terminate button should appeared in the last column.

To replace the terminated node, just simply upscale the previously terminated node's hostgroup with new nodes.

If the defective node is in a master hostgroup (that contains master components) then I am afraid there is no possible solution to replace the node through Cloudbreak and there is no easy way to replace it in Cloudbreak.

Br, Tamas

[#3]

Hi,

Is there a cbd shell command to terminate a unhealthy instance? Any help would be appreciated.

Thanks,

Cibi