Limits

Kupe Cloud’s limits split into two kinds: hard numbers you can plan against (token lifetimes, secret sizes, quotas), and softer constraints around timing and feature maturity. This page covers both, in the order most people hit them.

Cluster resource limits

Each cluster is created with explicit CPU, memory, and storage limits. Size the cluster for current workload needs, keep headroom for rollouts and retries, and review usage before resizing rather than waiting for pressure to show up in production. If a workload needs more room, raise the cluster’s limits; capacity is not open-ended by default.

Access and session limits

OIDC kubeconfig tokens last 8 hours.
Token kubeconfig lifetime is fixed at generation time.

Pick the shortest lifetime that gets the job done for automation, and rotate long-lived tokens regularly.

Secret limits

Each Secret object is capped at ≈1 MiB total (the Kubernetes API object-size limit). This bites when the secret is synced into a cluster, not when it is stored.
A secret can sync to up to 100 targets (cluster/namespace pairs), enforced by the platform’s API schema.

If you’re hitting the size limit, split large payloads across multiple secrets or move bulk configuration into regular config objects. Secrets are not a general storage layer.

Routing and certificate timing

DNS changes, certificate issuance, and GitOps reconciliation all take time rather than happening instantly. If a route or domain isn’t ready immediately, check propagation and reconciliation status before assuming the configuration is wrong.

Lifecycle guardrails

Create, upgrade, and delete are explicit operational events, not background details. Plan and validate upgrades like any other platform change; treat delete as destructive and irreversible. Lifecycle phases such as Provisioning, Running, and Upgrading are the signal to watch, for both your team and your automation.

Feature boundaries

Shared-node clusters are the standard self-service path today; dedicated node placement isn’t the default workflow yet. Some tasks are easier through the console, others through the API or Terraform. Build on the supported path first, then layer in more advanced automation once that’s stable.