Shared and team workers

A worker performs source processing and exports. Auto worker selection uses a compatible online team worker when available and can use shared cloud capacity allowed by the team tier. A project or job can also target a specific worker.

Open Workers in the Manage sidebar to see the workers available to the active team. Availability, creation rights, concurrency, and shared-worker access depend on the tier and your team role.

Create a self-hosted worker

On Workers, select New worker. STEP 1 Basics asks for Worker name, cache limit in GB, and a mount path. STEP 2 Settings sets Parallel tasks from 1 to 32 and can cap RAM per task and the percentage of CPU the worker may use. STEP 3 Deployment provides the command for the worker computer.

Review the settings and create the worker. Copy the generated deployment command for the operating system and shell you will use on the worker computer. Run that command on the intended machine with Docker available. The command contains a fresh enrollment token; do not post it in chat, screenshots, or public logs.

Return to the worker page and wait for Online. A newly created record does not process work until its container has connected successfully.

Read the worker list

Use search and status filters to find a worker. Online workers can accept work; Offline workers are not connected; Stopping means shutdown is in progress. Shared capacity is identified separately from workers owned by the active team.

Resource indicators summarize current CPU, memory, disk, and job use. A worker can be online but fully occupied, so compare running jobs with its configured concurrency before assuming a queued task is stuck.

Monitor a worker

Open a worker to see its status, work totals, running-job count, concurrency, RAM and CPU limits, live CPU/memory/disk readings, history charts, and recent jobs. Use this page when a task is slow, waiting, or failing only on one machine.

The import-cache panel shows used space and the configured limit. Use its cleanup action when cached imports occupy space you need. Cache cleanup removes reusable downloaded inputs, not project assets or completed exports; future jobs may need to download the data again.

Upgrade or redeploy

The worker image panel compares the current, desired, and latest available worker versions. Turn on Auto-upgrade when the machine should follow supported updates automatically. If the page asks for redeployment, copy a fresh deployment command and follow the shown replacement instructions on the worker computer.

Do not interrupt an active export just to apply a routine update. Wait for running work to finish, then redeploy and confirm that the worker returns Online before sending more jobs to it.

Change settings or remove a worker

Use Settings on the worker detail page to change the display name, concurrency, cache folder and limit, RAM-per-task cap, CPU limit, or auto-upgrade choice. Save, then watch the status to confirm the worker accepted the new configuration.

Remove permanently disconnects the worker record from the team. Stop or finish its active jobs first, confirm that you selected the right worker, remove it in MapToPlay, and then stop the corresponding container on that machine. Removing the worker does not delete project assets already produced.