Erupt Kubernetes Data Source

The erupt-data-k8s module provides a Kubernetes data source built on the fabric8 kubernetes-client. Bind an @Erupt model to a resource type (Pod, Deployment, Service, ConfigMap, Node, CustomResource, etc.) and you get a filterable, searchable, permission-controlled resource management view, with support for extracting arbitrary fields from the resource's nested spec / status.

Read-only + delete. Add and edit are not supported: building a K8s spec through a form is far less reliable than writing YAML directly.

Adding the Dependency

<dependency>
  <groupId>xyz.erupt</groupId>
  <artifactId>erupt-data-k8s</artifactId>
  <version>${erupt.version}</version>
</dependency>

The @EruptK8s Annotation (class level)

Attribute	Default	Description
apiVersion	—	API group / version, e.g. v1, apps/v1, batch/v1
kind	—	Resource type, e.g. Pod, Deployment, ConfigMap
namespace	""	Target namespace; empty means a cluster-scoped resource or all namespaces
masterUrl	""	Master URL override
kubeConfigPath	""	Path to a kubeconfig file
token	""	Bearer token override
maxItems	1000	Maximum number of items per list call

When kubeConfigPath / masterUrl / token are all empty, the standard discovery chain applies: KUBECONFIG environment variable → ~/.kube/config → in-cluster ServiceAccount.

The @EruptK8sField Annotation (field level)

Maps a model field to a JSON path inside the resource, e.g. spec.replicas, status.phase, spec.template.spec.containers[0].image. Dots separate object keys; [n] indexes into arrays.

The following field names resolve automatically without @EruptK8sField:

name, namespace, uid, resourceVersion, creationTimestamp, labels, annotations, kind, apiVersion, metadata, spec, status

metadata.generation is read into the resource data but is not in the shortcut list above — a field simply named generation resolves to null. Declare it explicitly with @EruptK8sField("metadata.generation").

Example 1: Deployment Dashboard

@Getter
@Setter
@Erupt(name = "Deployments", primaryKeyCol = "name")
@EruptK8s(
    apiVersion = "apps/v1",
    kind = "Deployment",
    namespace = "default",
    kubeConfigPath = "/Users/me/.kube/config"
)
@EruptDataProcessor(EruptK8sDataService.DATA_PROCESSOR)
public class K8sDeployment {

    @EruptField(views = @View(title = "Name"))
    private String name;

    @EruptField(views = @View(title = "Namespace"))
    private String namespace;

    @EruptK8sField("spec.replicas")
    @EruptField(views = @View(title = "Replicas"))
    private Integer replicas;

    @EruptK8sField("status.readyReplicas")
    @EruptField(views = @View(title = "Ready"))
    private Integer readyReplicas;

    @EruptK8sField("spec.template.spec.containers[0].image")
    @EruptField(views = @View(title = "Image"))
    private String image;

    @EruptField(views = @View(title = "Created At"))
    private String creationTimestamp;
}

Example 2: Pod Status Board

@Getter
@Setter
@Erupt(name = "Pods", primaryKeyCol = "name")
@EruptK8s(apiVersion = "v1", kind = "Pod", namespace = "prod")
@EruptDataProcessor(EruptK8sDataService.DATA_PROCESSOR)
public class K8sPod {

    @EruptField(views = @View(title = "Name"))
    private String name;

    @EruptK8sField("status.phase")
    @EruptField(views = @View(title = "Phase"), edit = @Edit(search = @Search))
    private String phase;

    @EruptK8sField("spec.nodeName")
    @EruptField(views = @View(title = "Node"))
    private String node;

    @EruptK8sField("status.containerStatuses[0].restartCount")
    @EruptField(views = @View(title = "Restarts"))
    private Integer restarts;

    @EruptK8sField("status.podIP")
    @EruptField(views = @View(title = "IP"))
    private String podIp;
}

Supported Operations

• List: fetches all resources of the type (capped by maxItems), then filters / sorts / paginates in memory
• Detail: withName(id).get(), using the resource name as the primary key
• Delete: withName(id).delete()
• Add / Edit: not supported

Note

• The primary key value is the resource name; the model's primary key column should map to metadata.name (or simply be a field named name).
• Leave namespace empty for cluster-scoped resources (Node, PersistentVolume, ClusterRole, etc.).
• Results beyond maxItems are silently truncated — if you hit the cap regularly, raise the value or split the model by namespace.
• maxItems truncates client-side only: fabric8's list() still pulls every resource of that type back to the JVM before the first N are kept. Lowering it saves memory, but not the cost of the API call.

Read-only means "errors on submit", not "buttons hidden"

addData / editData throw outright, but the service does not override power() — no data source under erupt-data does. So the Add and Edit buttons still render in the admin list, and the user only sees the error after filling in the form and hitting submit.

Turn those two permissions off explicitly on the model:

@Erupt(
    name = "Deployments",
    primaryKeyCol = "name",
    power = @Power(add = false, edit = false)
)

Add delete = false as well if you want to block deletes too.

Contributors

YuePeng

Claude Fable 5.1

Changelog

Last edited 4 minutes ago

View full history

9d3089d-Split the erupt-tpl module docs and drop the obsolete memory screenshot (#15)