Support rollback for Kubernetes workloads that do not create long-running instances, such as Jobs and CronJobs.
Problem
Harness rollback relies on tracked instances. Jobs and CronJobs are short-lived and do not register instances, preventing rollback even when manifest changes need to be reverted.
Requested Capability
  • Rollback based on manifest or release history rather than instance tracking
  • Support declarative rollback for short-lived workloads
  • Visibility into previous successful executions for recovery
Value
Provides safe recovery for batch and scheduled workloads, which are common in data processing and financial systems.
Created by Gino Orvieto
·