Applications
Applications on Kupe Cloud run as standard Kubernetes workloads — Deployments, StatefulSets, Jobs, and CronJobs. You package your application as a container image, define its runtime requirements in Kubernetes manifests, and deliver it through GitOps.
Deployment model
Section titled “Deployment model”Kupe Cloud uses a GitOps-first delivery model powered by Argo CD:
- Define — write Kubernetes manifests, Helm charts, or Kustomize overlays and commit them to Git.
- Sync — Argo CD watches your repository and applies changes to your cluster.
- Validate — health checks, readiness probes, and observability confirm the rollout is healthy.
- Promote — move changes across environments (dev → staging → production) via Git.
There is no CI/CD pipeline pushing to the cluster. Git is the source of truth, and Argo CD pulls from it.
What a production workload needs
Section titled “What a production workload needs”Before going live, every workload should have:
- Health probes — liveness and readiness probes so Kubernetes can manage pod lifecycle.
- Resource requests and limits — CPU and memory bounds for scheduling and cost visibility.
- Service and route definitions — an
HTTPRoutefor external traffic, aServicefor internal communication. - Manifests in Git — everything versioned and deployable via Argo CD.
- Observability — metrics scrape annotations, structured logs, and alert rules for key SLOs.
Supported workload types
Section titled “Supported workload types”| Resource | Use case |
|---|---|
| Deployment | Stateless services (APIs, web servers, workers) |
| StatefulSet | Stateful services (databases, caches with persistent identity) |
| Job | One-off tasks (migrations, data processing) |
| CronJob | Scheduled tasks (reports, cleanup, backups) |
Guides
Section titled “Guides”- Deploy Applications — step-by-step deployment workflow with examples.
- GitOps Workflows — promotion, rollback, and drift management patterns.