The scenario
A client submits a task to a scheduler. The scheduler queues the task; a worker processes it. On failure, the scheduler retries according to retry policy, rescheduling the task up to a maximum retries before marking it failed.
What is in this drawing
Read the decisions behind it.
01
Implement retry logic with max retries and backoff
02
Use a task queue to decouple submission from processing
03
Notify client on completion or failure
When designing a distributed task scheduling system with fault tolerance and retries, such as job queues, cron replacements, or workflow engines.
More like this