Summary:

Jira uses its internal coordination mechanism, Jira employs Hazelcast for cluster management, to elect a node responsible for creating the index snapshot. This is often the "active" node or a node that has been designated for such tasks based on the current load and resource availability. 

Example:

Our datacenter runs in AWS and have scheduled autoscaling to scale down the cluster after business hours for cost efficiency reasons.

The node which is randomly picked by Jira for index snapshot creation could be terminated as a part of Scale down process causing the the index snapshot creation process interruption.

This will lead to corrupt index snapshot creation and will impact the Jira installation where Jira will not be able to recover its index from corrupt snapshot. 

Suggestion:

We would like a feature in Jira where Jira Admins could choose the node in the cluster to run the index snapshot creation operation.