AWS Step Functions State Machine
Does ZopNight manage AWS Step Functions State Machine?
AWS Step Functions bills Standard workflows per state transition and Express workflows per request plus GB-second duration, so a looping state machine meters on every iteration. ZopNight tracks execution cost per state machine and can halt in-flight executions via states:StopExecution during scheduled off-hours windows.
Rules that fire on AWS Step Functions State Machine
At a glance
| Field | Value |
|---|---|
| Stop verb | states:StopExecution |
| Scheduling notes | in-flight executions are halted via states:StopExecution during scheduled off-hours windows. |
AWS Step Functions orchestrates multi-step workflows across AWS services, billed per state transition (Standard) or per request and duration (Express). Long-running or looping executions can quietly drive both Step Functions charges and downstream compute cost.
Standard counts transitions; Express counts time
A Standard workflow bills per state transition: every step of every execution, retries included, is a metered event. An Express workflow bills per request plus GB-seconds of duration, like Lambda. The same definition can be dramatically cheaper on one type than the other depending on its shape: high-volume, short-lived workflows favor Express, while long-running or human-in-the-loop workflows require Standard and its per-transition meter.
Loops multiply the meter
The cost amplifier in Standard workflows is iteration. A Map state fanning out over 10,000 items, or a polling loop of wait, check, repeat, turns one execution into thousands of billable transitions, and each transition may also invoke Lambda functions or ECS tasks that bill on their own meters. Generous retry policies do the same thing in slow motion, re-billing failed paths for hours.
What stop means for a state machine
A state machine definition has no stopped state; it is configuration, permanently ready to execute. Scheduling therefore acts on executions: during a scheduled off-hours window, ZopNight halts in-flight executions via states:StopExecution. This fits non-production machines whose executions arrive from schedules and queues that should stay quiet overnight; halted executions terminate rather than pause, so anything mid-flight is cut, not resumed.
Coverage and cost
State machines are discovered on the 6-hour cycle, with execution cost attributed from Cost Explorer or CUR 2.0 and usage-based recommendations built on execution volume.
Executions in the console
Step Functions console, then State machines, then a machine’s Executions tab. A machine with executions perpetually in Running status is where transition counts and downstream compute accumulate unseen. Sorting that list by start time is worth doing whenever a workflow bill surprises.