- About this guide
- Getting started
- Installing
- Upgrading
- Connecting and licensing
- Configuring
- Running workloads
- RPA processes and tests
- Agents, functions, and API workflows
- Limits
- Securing and governing
- Monitoring
- Troubleshooting
- Reference
How a job runs
Job lifecycle between Orchestrator and UiPath Robot, from start to dispatch to completion, and how runtimes and slots decide where and when a job can run.
A job is one execution of a process. Orchestrator receives it, picks a robot that can run it, and hands it over. The Robot runs it and reports back. The path is the same for every workload. What changes is the capacity Orchestrator checks before it hands the job over.
Agents, functions, and API workflows run only with Automation Cloud or Automation Suite Orchestrator, on robots connected with client credentials.
From start to finish
- A job starts for a process: manually, from a schedule or a queue trigger, through the API, or from another automation.
- Orchestrator reads the process type and the runtime type the job asks for.
- Orchestrator looks for capacity on the machine objects in the job's folder: a free runtime for an RPA or test job, or a free slot for an agent, function, or API workflow job.
- Connected robots report which process types and runtime types they support. Orchestrator dispatches the job only to a robot that supports both.
- Unified Runtime hands the job to the executor for its workload: the .NET executor for RPA processes and test cases (Windows - Legacy projects run in the .NET Framework executor), an executor package for agents, functions, and API workflows.
- The Robot sends status, logs, and traces back to Orchestrator while the job runs.
- When the job ends, the runtime or slot it used becomes free again.
Runtimes and slots
Orchestrator measures capacity in two ways. Which one applies depends on the workload.
| Property | Runtimes | Agent slots and Function slots |
|---|---|---|
| Used by | RPA processes and test cases. Also functions and API workflows, when Use Production (Unattended) is selected on the machine template. | Agents use Agent slots. Functions and API workflows use Function slots. |
| What one unit allows | One concurrent job on the machine | One concurrent job of that type on each connected machine |
| Set on | The machine object or machine template | The machine template, in Agents and Functions runtimes |
| Limited by | The runtimes allocated to the tenant | Nothing at license level. Agent jobs consume units. Function and API workflow jobs consume an unattended runtime or units. |
| Default | Set when you create the template | 0 Agent slots and 0 Function slots |
Runtimes come from the tenant's pool. A machine takes the runtimes of its machine object when it connects to Orchestrator, and returns them when it disconnects. Slots reserve execution capacity on the machine itself. They are not drawn from a license pool, but a job still needs units to start, or, for a function or API workflow, an unattended runtime.
Foreground processes
A foreground process uses the screen, so a Windows user can run only one foreground job at a time, and only one foreground job can run in the console session. If a second foreground job starts for the same user, the Robot doesn't queue it. The job fails with the error "Another interactive job is running for this user".
On a Windows Server machine, several users can each run one foreground job in their own RDP session. Background processes are not limited this way. This applies to Windows workstations as well as Windows Server: how many background jobs run at the same time depends on the runtimes, licenses, or units available. For details, refer to Background automations.
When no robot can take the job
Orchestrator treats the two kinds of capacity differently when nothing is available:
- RPA processes and test cases: if no matching runtime is available, Orchestrator doesn't let the job start.
- Agents, functions, and API workflows: the job stays Pending until a robot that supports its process type connects. It doesn't fail.
- No units left: an agent job doesn't start, even when slots are free. Neither does a function or API workflow job that runs on units.
Related topics
- Managing machines, to set runtimes and slots on a machine template
- Agents and Functions on Local Robots, for the Local runtime type
- Job executions, for how the Robot runs jobs in Windows sessions