Rancher cluster id (for example local or c-m-xxxx). The plugin uses it in GET /v3/clusters/{id}, manifest apply, and Helm catalog calls.

In Rancher Manager: open the cluster, then check the browser URL (…/dashboard/c/<cluster-id>/…) or Cluster → Details (id field). For the local cluster, the id is often local.

SUSE KB: How to retrieve Rancher cluster ID

Helm release name in the target namespace (DNS-1123).

Chart name within the repository (for example nginx).

Chart repository URL that matches an existing Rancher ClusterRepo spec.url.

Supported schemes:

• https://charts.example/helm — classic Helm index repository
• http://charts.example/helm — same (HTTP)
• oci://registry.example/charts — OCI chart registry (Rancher Manager 2.9+)

If catalog cannot find the chart version, the step refreshes that ClusterRepo like the Rancher UI (sets spec.forceUpdate and waits) and retries once. The ClusterRepo must already exist in Rancher; the step does not create it. Registry credentials for private OCI hosts belong on the ClusterRepo in Rancher, not on this step.

Adds wait and cleanupOnFail to the Rancher catalog install or upgrade body (Helm --wait / rollback on fail when the API supports it).

This does not change the Jenkins wait: the step still polls the Helm app and then release workloads. A Helm rollback does not make the build succeed. Catalog HTTP 200 is not success by itself.

When enabled, create the namespace in the selected Project if it does not exist. On a newly created namespace, the step also applies a default ResourceQuota (namespace-quota) and LimitRange (namespace-limits) to cap pods and CPU/memory.

When disabled, the namespace must already exist in that project. Existing namespaces are never moved or patched. Skipped when Validate only is checked.

When checked, uninstall the release (if present) then install.

Destructive — prefer the default install-or-upgrade path unless you need a clean reinstall.

Target Kubernetes namespace (Helm --namespace). Default default.

The namespace must belong to the Project field. When Ensure namespace is enabled and the namespace is missing, the plugin creates it in that project. An existing namespace in a different project (or in none) fails the build; the step does not move namespaces.

Rancher project name as shown in the UI (for example Default), or the project id (p-xxxxx). The plugin lists projects with GET /v3/clusters/{clusterId}/projects for this step’s Cluster ID.

Do not put the cluster id here (not local:p-xxxxx). Cluster belongs in Cluster ID.

Inherit from System: This setting is made in the System tab of Manage Jenkins (Rancher Manager).

Manual: set Rancher URL and Secret text API token on this step.

Secret text with the Rancher Bearer Token (token-…:… from Account API Keys), not Access Key alone.

Rancher Manager base URL (e.g. https://rancher.example).

Preflight and field validation only — no project lookup, namespace create, Helm install, upgrade, or uninstall.

Inline Helm values YAML when Values source is Manual YAML. Must be a mapping (key: value). Build logs do not include the YAML body.

Path to the values file relative to the repository root. Default values.yaml.

Optional Username with password or Secret text for private Git repositories used to fetch the values file. Secret text is used as a token with username oauth2 (common for GitLab/GitHub PAT).

Git reference for the values repository. Default: refs/heads/main.

Examples:

• refs/heads/main
• refs/tags/v1.2.0
• refs/heads/${BRANCH_NAME}

Git repository URL hosting the values file (not the chart repo).

Do not put credentials in the URL (no user:pass@host). Use Git credentials for private repos.

Example: https://gitlab.example/group/helm-values.git

Helm chart values: none (chart defaults), Git repository file, or inline YAML. Values are sent to Rancher catalog install/upgrade as JSON (never logged).

When enabled, debug-level logs for HTTP requests will be duplicated in the build console.

Optional chart version. Empty uses the repository default or latest resolved by Rancher.

How long Jenkins waits after catalog install or upgrade for this Helm operation to settle. First the named catalog operation (chartActionOutput → GET catalog.cattle.io.operations metadata.state) must leave Steve transitioning without error. Then the Helm app summary.state (including pending-upgrade / pending-rollback) and helm-owned objects on the app (relationships). Default 300 seconds. Must be greater than zero (empty uses the default). The same timeout also bounds ClusterRepo index refresh when catalog reports the chart is missing.

A leftover failed or deployed summary from before this catalog call is not the result yet. A failed operation (helm-pod exit not 0) fails the build immediately with that kstatus message and the tail of the Helm job log (Steve operation /logs, same as Recent Operations → Logs). A long log is truncated so exit code: N and the ending Error: / FAILED stay visible. An empty 2xx log body is marked helm job log was empty (not a bare exit code alone). 404 leaves only kstatus; other log HTTP errors add a short cannot-read note with status. Jenkins does not wait on a missing app. Settled app failed fails the build even if workloads still look active (cluster serving). Settled ready with a Deployment/StatefulSet that never becomes active also fails the build. HTTP GET paths are debug (Verbose logging). Console INFO is state and object changes of this operation.

Rancher cluster id (for example local or c-m-xxxx). The plugin uses it in GET /v3/clusters/{id}, manifest apply, and Helm catalog calls.

In Rancher Manager: open the cluster, then check the browser URL (…/dashboard/c/<cluster-id>/…) or Cluster → Details (id field). For the local cluster, the id is often local.

SUSE KB: How to retrieve Rancher cluster ID

Optional Username with password or Secret text for private Git repositories used to fetch the manifest file. Secret text is used as a token with username oauth2 (common for GitLab/GitHub PAT).

Path to the manifest YAML inside the repository (relative; default manifest.yaml).

Repository: Jenkins fetches YAML from Git and applies it via Rancher (Steve action=apply).

Manual YAML: paste manifest content in the job configuration.

Example repository: https://gitlab.example/group/manifests.git

Inline Kubernetes YAML when Manifest source is Manual YAML. Applied via Rancher Steve action=apply. Namespace comes from the YAML. Build logs do not include the YAML body.

Inherit from System: This setting is made in the System tab of Manage Jenkins (Rancher Manager).

Manual: set Rancher URL and Secret text API token on this step.

Secret text with the Rancher Bearer Token (token-…:… from Account API Keys), not Access Key alone.

Rancher Manager base URL (e.g. https://rancher.example).

Git reference for the manifest repository. Default: refs/heads/main.

Examples:

• refs/heads/main
• refs/tags/v1.2.0
• refs/heads/${BRANCH_NAME}

Git repository URL hosting the manifest file.

Do not put credentials in the URL (no user:pass@host). Use Git credentials for private repos.

Example: https://gitlab.example/group/manifests.git

Resolve connection and validate manifest source without calling Rancher apply or namespace create.

When enabled, debug-level logs for HTTP requests will be duplicated in the build console.

How long Jenkins waits after Steve apply for Deployment, StatefulSet, DaemonSet, and Job resources declared in this YAML to become ready (Steve metadata.state active or Kubernetes ready replicas / Job completions). Default 300 seconds. Must be greater than zero (empty uses the default).

Manifests with only ConfigMap, Service, Namespace, and similar kinds succeed after apply without polling. Apply HTTP OK alone is not success when those workloads are present. On timeout, the build fails with the last workload states and one pods list hint (not on every poll tick). Validate only skips apply and wait.