Propose a high-level architecture for a distributed cron-like system, identifying the main components (e.g., scheduler, worker, storage, coordination service) and their interactions.

Detail how a distributed cron-like system would schedule complex, time-based jobs. This includes supporting various cron-like expressions (e.g., 'every 5 minutes', 'every Monday at 9 AM'), one-time jobs, and recurring jobs. Discuss how to handle jobs that might be delayed or skipped.

Explain how to ensure a job runs only once, even with multiple potential schedulers or workers in the cluster. Describe the mechanism for leader election and how it prevents duplicate job executions.

Describe how the system handles the failure of individual machines (e.g., a scheduler node, a worker node, or a storage node). Focus on fault tolerance, high availability, and how jobs are retried or reassigned without loss or duplication.

Elaborate on how job definitions (scheduling rules, commands to execute, metadata) are stored persistently. Discuss the choice of storage technology (e.g., relational database, NoSQL, distributed key-value store) and considerations for consistency, durability, and availability.

Discuss how the system can scale horizontally to handle a large number of jobs and a large cluster of machines. Mention potential bottlenecks and how to mitigate them.

Briefly describe how users would define, manage, and monitor jobs within this distributed system.