items.automations.title

Automations let you run recurring work on a schedule without deploying a long-lived application. Each automation runs as a real Kubernetes CronJob in your organization's environment, executed through the nebua-agent.

Overview#

Instead of writing raw manifests, you pick an automation type and fill in a short form. NebuaCloud builds the CronJob, schedules it, captures each run's logs, and can alert you when a run fails.

Automation types#

Type	What it does
Health check	Periodically probes a URL or a linked app and alerts when it stops responding or returns an unexpected status.
Script	Runs an inline shell / Node / Python script, or a command in a container image (including your app's own image).
Webhook	Calls an external API or webhook on a schedule, with custom method, headers and body. Alerts on non-2xx responses.
Security check	Runs a scheduled scan (Trivy image/filesystem, npm audit, kube-bench) and fails — and alerts — past a severity threshold.
AI task	Runs scheduled AI processing over recent logs, deploy events or metrics, and stores or emails the result.
Custom job	Start from scratch: your own image, command, arguments, environment, schedule and resource limits.

Custom jobs and existing CronJobs#

Any CronJob that already exists in your environment namespace — created with kubectl, Helm, or another tool — is listed here as a Custom Job with an external badge. You can pause, resume, run, inspect and delete it. Editing and saving it "adopts" it: NebuaCloud adds its own labels so it becomes a fully managed automation.

Creating an automation#

1. Open Automations in your dashboard.
2. Click New automation and choose a type.
3. Fill in the type-specific fields (target URL, script, webhook URL, prompt, image, …).
4. Set a schedule — pick a preset or enter a standard 5-field cron expression. The form shows the next run time as you type.
5. Optionally add a notification email and choose to be alerted on failure.
6. Set advanced options if needed: timeout, retries, enabled/paused.
7. Create.

Schedule formats#

Automations use standard cron expressions (UTC by default):

• */15 * * * * — every 15 minutes
• 0 * * * * — every hour
• 0 2 * * * — every day at 02:00
• 0 3 * * 0 — every Sunday at 03:00

Run history#

Every automation keeps its recent runs (successful and failed), each with start/end time, duration, status and the pod logs. Use Run now to trigger a run immediately without waiting for the schedule.

Notifications#

When you set a notification email and enable alert on failure, NebuaCloud sends an email and an in-app notification when a run fails.

Best practices#

• Keep each automation focused on one task.
• Prefer idempotent commands so a re-run is safe.
• Store credentials as environment variables / secrets, not in plain command arguments.
• Set a timeout so a stuck run can't run forever.
• Start paused or use Run now to test before relying on the schedule.

Troubleshooting#

The automation never runs — check the cron expression is valid (the form flags invalid ones), confirm it isn't paused, and that a nebua-agent is connected for your cluster.

Runs fail immediately — open the run in Run history and read the pod logs. Common causes: a bad image reference, a missing environment variable, or an unreachable target.