Managed Machines
Trickest-provided compute with automatic scaling.
On this page5
Overview
Managed machines are virtual machines that Trickest provisions and operates for you. You run workflows on a managed fleet without installing, configuring, or maintaining anything — the platform handles machine lifecycle entirely.
Managed fleets are available on Enterprise plans. On the Community plan, workflows run on self-hosted machines.
How Managed Fleets Work
Your workspace gets a managed fleet with a machine cap defined by your subscription. Machines in it are ephemeral: Trickest starts them when runs need capacity and shuts them down when they are no longer needed. You don't manage individual machines — there is nothing to patch, restart, or resize.
Scaling
Managed fleets are autoscaled by Trickest. Machines are never allocated in advance — they are started when jobs need them and turned off as soon as they don't:
- When a run starts, the platform starts only as many machines as there are jobs ready to execute. If the workflow begins with 3–4 nodes in parallel, 3–4 machines come up — not the run's maximum.
- As the run progresses and more jobs become ready (fan-outs, distributed nodes), the platform starts additional machines to match, up to the fleet's capacity.
- As soon as a machine has no more work, it is turned off, freeing the slot for other jobs or runs.
- You don't size the run yourself; the machine count shown when executing is the run's upper bound, and you can lower it to cap the run's footprint.
Fleet Limits and Sharing
The fleet's machine cap is a hard limit. A single run never exceeds it — thousands of jobs can be queued while the run uses only the machines available to it.
Multiple runs can be active on the same managed fleet at the same time. They share the fleet's pool, and machines are assigned to whichever run has pending work as they become free. There is no manual reservation.