kunobi-ninja/Kobe Cluster

Claim a Kubernetes cluster from a Kobe pool. Auto-releases on job exit.

View on GitHub

Trust Signals

Scorecard Score
not yet scored
Maintenance Recency
Stale
License
None
namedescriptionrequireddefault
endpointKobe API endpoint URLyes
poolPool name to claim from (e.g. ci-small)yes
ttlLease TTL (e.g. 1h, 30m)no1h
audienceOIDC token audiencenokobe-system
timeoutMax time to wait for the lease to become Bound (e.g. 5m)no5m
wait-for-readyAfter lease is Bound, also wait until cluster nodes report Ready. Defaults to false for backward compatibility — set true to block until the cluster is fully usable for tests.nofalse
min-ready-nodesMinimum non-control-plane nodes that must be Ready (only used when wait-for-ready=true). Default 0 — single-node k3s/k0s server pools (the common CI shape) pass without further config since the server itself is labeled control-plane. Set to 1 or higher for multi-node setups that require a worker.no0
ready-timeoutMax time to wait for cluster readiness after Bound (only used when wait-for-ready=true). Independent from `timeout`, which gates the Bound transition.no2m
namedescription
kubeconfig-pathPath to the kubeconfig file
lease-idLease ID
cluster-nameName of the claimed cluster
cluster-backendAuto-detected cluster backend: k3s | k0s | kubernetes | unknown. Always set when kubectl is on PATH; useful for backend-aware downstream steps. The `kubernetes` value is a generic bucket covering vanilla, kind, capi-managed, EKS/GKE/AKS, and vcluster (vcluster reports the underlying distro's gitVersion — typically k3s).