Limits and quotas

Resource reservations

When you create a container in the portal you reserve CPU, memory, and disk. These are bounded and step-locked. The create and edit dialogs only accept values in range:

Replicas are not set when you create a container (they default to 1). You scale them later in the container's Edit dialog. If autoscaling is enabled, the autoscaler owns the replica count and manual changes are rejected with AUTOSCALING_OWNS_REPLICAS. Out-of-range values are rejected with BAD_RESOURCES.

What you reserve is billed pay-as-you-go from your balance. See vouchers and billing for how usage is metered.

Container model

• Port is fixed at 8080. There is no port field when you create a container; the platform always routes to 8080. Your image must listen on 0.0.0.0:8080 (see container requirements).
• No image or domain field at create time. A container boots on a managed placeholder image; you ship your real image later via GitHub Actions. A Public container is auto-assigned a *.alawadi.cloud hostname with managed TLS; an Internal container is reachable only by cluster DNS inside the project.
• Environment variable keys must match ^[A-Z_][A-Z0-9_]*$ (uppercase letters, digits, and underscores). Duplicate keys are rejected.

Project quotas and isolation

• Resource quotas (CPU, memory) cover every service in the project namespace.
• Limit ranges apply default and maximum CPU/memory to each container.
• Workloads run under restricted Pod Security: no privileged mode, host networking, hostPath, or Docker socket.

AI rate limits

The built-in AI is shared, so requests run under fixed rate limits: at most 60 requests per minute and 1,000 per hour, with no more than 4 concurrent requests in flight at once. Each request uses a 16k-token context window. If AI is turned off for your account, the API returns 503 AI_DISABLED.