Part 1 of 8 · 1 min read
Why this page comes seventh in Level 13
In this module you will
- Scope a scheduler for periodic jobs and say which failures it must avoid.
- Explain why a single-machine cron is a single point of failure.
- Design a leader and followers with a replicated log of launches.
- Make launches safe with two sync points, idempotent job names and launch times.
- Handle log growth, snapshots, thundering herds and missed runs.
Almost every company runs jobs on a timer: clean-ups, reports, billing runs, backups. A scheduler looks like a clock and a loop, yet a distributed one forces the hardest questions of Level 5: who is the leader, what if two think they are, and what if a crash comes in the middle of a launch? The page draws on the SRE chapter "Distributed Periodic Scheduling with Cron" (Site Reliability Engineering, Google). The ideas of leaders, replicated logs and idempotency come from Levels 5 and 8 and from pages 9 and 10 of this level. Numbers marked "made up" only show the idea.