Final answer:
During a rolling-restart, in-progress searches are usually handed over to other operational nodes in the cluster, minimizing disruption. However, this depends on the system's capabilities and configuration; if not supported, in-progress searches may fail and need to be restarted.
Step-by-step explanation:
When a rolling-restart is initiated on a system that handles searches, such as a search engine or a database cluster, it is designed to minimize downtime and service disruption. The goal of a rolling restart is to update or restart servers one by one, without taking the entire cluster down at once. As a result, in-progress searches are typically not terminated but are handed off to other nodes in the cluster that are still operational. If the cluster's architecture supports such a feature, it will ensure that any search in progress at the time one node goes down for a restart will be taken over by another node. However, this behavior can vary depending on the specific implementation and configuration of the system. If the system doesn't support handing off searches, in-progress searches on the node being restarted might fail and would need to be restarted once the rolling-restart is complete.