Skip to main content
By default every deployment runs on slipway’s infrastructure with nothing to configure. Bringing your own cluster runs them on yours instead, so workloads, images, logs and secrets stay inside your network. slipway reaches your cluster through an agent you install with Helm. The agent connects outbound to the control plane over a single secure connection and does the work from inside your cluster: builds, deploys, ingress, snapshots, logs and shells.
slipway never receives a kubeconfig and never holds your cluster credentials. The only secret it stores is the agent’s outbound auth token, hashed. Because the connection is outbound only, private and firewalled clusters work with no inbound access.
Registering a new cluster requires the Pro plan or higher. Clusters you have already registered keep working if the plan changes; only registering new ones is gated. See Plans.

​
Prerequisites

Your cluster needs these already installed. The slipway chart does not add them.

​
Setting it up

  1. Register the cluster on the Clusters page. slipway mints a one-time agent token and returns the exact helm command to run. The token is shown once.
  2. Install the agent with that command. It installs the build tooling, an in-cluster registry and the agent. The cluster flips from pending to ready once the agent’s first connection lands.
  3. Point a project at it, or set it as your organization’s default target. New deployments for that project then build and run in your cluster.

​
Registry authentication

The chart’s in-cluster registry accepts unauthenticated pulls from inside your cluster by default, which is usually fine because the cluster is yours alone. To require authentication, set registry.auth.tokenCert in the chart values to the public certificate of slipway’s registry token signer, available from support. The registry then serves only clients carrying tokens minted by the control plane, scoped to your organization’s images.

​
Domains

slipway assigns each cluster a subdomain under its own zone and provisions DNS and TLS for you, with nothing to set up. To serve from a domain you own, register the cluster with your own apps domain, or claim the domain under Domains and bind it to the cluster. slipway then shows the DNS record to add. A domain is bound to one cluster at a time.

​
Billing

Workloads on your own cluster are still metered, at a reduced rate set per cluster. You supply the hardware; slipway supplies the control plane, the build pipeline, log and metric ingest, and support. The rate is shown on the cluster’s page. Disks on your cluster live on your storage and are not billed by slipway. See Billing.

​
Limitations

  • Projects on your cluster are not sampled for metrics, so their Metrics tab is empty.
  • Disk usage additionally needs an agent recent enough to report volume statistics.
  • In-place compute resize needs the pods/resize RBAC rule. An agent predating it falls back to advising a redeploy. See CPU and memory.