DataStax Help Center

Lifecycle Manager: auto_bootstrap null after running configure on an imported cluster

Introduction

One of the first actions you are likely to take after importing an existing cluster into LCM is to change some configuration and push it out to your cluster via a "Configure" job. Unfortunately, there is a known issue in LCM (currently affects versions of OpsCenter from 6.0.0-6.0.2) that causes it to set auto_bootstrap to null in cassandra.yaml. This prevents DSE from starting back up.

Solution

The solution to this problem is easy. Simply run an Install job instead of a Configure job. Install jobs are implemented to be idempotent - this means that running an Install job multiple times with no configuration changes should result in the same cluster "state". Since DSE is already installed on the nodes, it will not actually re-install anything. This means that Install will behave more like Configure in this scenario with the following small exceptions:

  • An Install job allows you to specify the value for auto_bootstrap prior to starting the job. Set this to False. This solves the problem.
  • If you only have 1 seed node in each DC, or less than 3 seeds total in your cluster, Install will pick one or more additional seed nodes such that you have at least 2 per DC and at least 3 total in the cluster.

WARNING: if you are using vnodes, make sure that num_tokens is set properly in the cassandra.yaml Configuration Profile for your cluster. This is important whether running Install or Configure, as Configuration Profiles default to a non-vnode setup.

The internal ticket number for the proper fix to cluster import is OPSC-10151.

Was this article helpful?
1 out of 1 found this helpful
Have more questions? Submit a request

Comments

Powered by Zendesk