AI Platform
Run a container image as a one-off job, without deploying an app. Start it from the dashboard or the API, read the logs, done.
# define the job once: image, command, size, region curl -X POST https://api.nodion.com/v2/runners \ -H "Authorization: Bearer $NODION_TOKEN" \ -H "X-Nodion-Project: $NODION_PROJECT" \ -H "Content-Type: application/json" \ -d '{ "name": "migrate", "image": "ghcr.io/acme/shop:v2", "run_command": "bin/rails db:migrate", "region_id": "reg_…", "timeout_seconds": 900 }'
# start a run; command and argv can be overridden per run curl -X POST https://api.nodion.com/v2/runners/$RUNNER_ID/runs \ -H "Authorization: Bearer $NODION_TOKEN" \ -H "Content-Type: application/json" \ -d '{"command": "bin/rails db:migrate:status"}' # then read status, exit code and logs curl https://api.nodion.com/v2/runners/$RUNNER_ID/runs/$RUN_ID/logs \ -H "Authorization: Bearer $NODION_TOKEN"
How it works
Define it once, run it as often as you need. Nothing keeps running between runs.
Built in
Every run starts in its own short-lived microVM and is torn down when it ends. Nothing leaks between runs.
Runs are network-isolated from your other services by default.
Pick a package for memory, CPU and disk, just like for an application process.
One hour by default. Long builds and batch jobs can run up to 24 hours.
Every run keeps its logs, status and exit code, in the dashboard and over the API.
Stop a run that is stuck or no longer needed with one call.
Use cases
Run the migration as its own job before or after a release.
Tests, builds and checks in a clean environment every time.
Kick off work from a webhook without keeping a worker running.
Give a coding agent a disposable place to run code and shell.
Earn up to 15 € in credit when signing up.
or with email