Process Instance Lifecycle
Every run of a process is one instance with a beginning, a middle, and exactly one end. The lifecycle is that arc: how a run is started, what it is doing while it lasts, and the state it comes to rest in.
How a run begins
- A person starts it. The start button on the Processes list launches an instance. If the process needs inputs, you fill them in on a start task first.
- An event starts it. A verified inbound webhook delivery starts or advances the processes subscribed to it.
- Another process starts it. A subprocess step launches a run of another process and passes values into it.
Whichever way it began, the instance gets its own copy of the process data and moves independently of every other run of the same process.
While it is running
The instance walks its path through the steps (see Control Flow), gathering data as it goes. Human steps appear as tasks and disappear again when they are submitted. The instance is pinned to the process revision it started on, so a later edit to the process does not change a run already in flight - you migrate it deliberately if you want it on a newer revision.
The states
| State | What it means | How it gets there |
|---|---|---|
Running | Advancing, or waiting on a task. | The state every instance starts in. |
Success | Reached an end step. Nothing is left to do. | The run completed on its own. |
Error | Halted on a failure nobody handled. | An unhandled error; the message names the step. |
Cancelled | Stopped deliberately before finishing. | Cancel Instance, by an authorized actor. Open tasks are removed. |
Paused | Suspended, and resumable. | The pause endpoint of the API; the run picks up exactly where it stopped. |
Success, Error and Cancelled are terminal - a run in one of them never
moves again. Paused is the only state a run returns from.
The instance list filters use everyday words for the same states: Ongoing, Done, Failed, Cancelled, Paused.
After it ends
Nothing is deleted. A finished instance keeps its data and its execution history, which is what makes a run auditable long after it closed. Open any of them from the process’s instance list, or from Monitor if you are an admin.
Related
- Control Flow - how the run moves from step to step in between.
- Process Instance View - where you watch and act on a single run.
- Error Handling - keeping a failure off the
Errorend.