Support Questions
Find answers, ask questions, and share your expertise

Unable to scale down nodes due to replication factor

Explorer

I have an HDP cluster, and I am trying to scale down some of it's nodes using cloudbreak. The specific group in question was created with 1 node, scaled up to 4 nodes, and now when I try to scale down I manage to scale the group down to 3 nodes but anything smaller than that gives me the following error -

`7/25/2016 1:35:07 PM horton-cluster-multi-roles - update failed: New node(s) could not be removed to the cluster. Reason There is not enough node to downscale. Check the replication factor and the ApplicationMaster occupation.`

The group host contains only the following components - node_manager, metrics_monitor. So I don't understand how replication factor is a consideration for these nodes.

1 ACCEPTED SOLUTION

Accepted Solutions

Expert Contributor

The default replication factor is 3 so we don't allow scaling a host group below 3. It's a known, bug that we don't consider the services installed on those hosts as in your case it does not affect the replication. It will be fixed in the following versions.

View solution in original post

5 REPLIES 5

Expert Contributor

The default replication factor is 3 so we don't allow scaling a host group below 3. It's a known, bug that we don't consider the services installed on those hosts as in your case it does not affect the replication. It will be fixed in the following versions.

View solution in original post

Explorer

Is there any way I could be informed when this bug is solved? Some kind of newsletter or release notes I can subscribe to?

Mentor

You can click watch this jira on the associated page. You need to find out the name of the Jira.

Expert Contributor

Cloudbreak is not an Apache project, yet so it's an internal JIRA, but I'll update this thread.

Expert Contributor

The fix went in to the rc-1.4 branch so the next release will solve this problem: https://github.com/sequenceiq/cloudbreak/pull/1814