Skip to content

Latest commit

 

History

History
189 lines (160 loc) · 5.52 KB

inventory_object.md

File metadata and controls

189 lines (160 loc) · 5.52 KB
Raw

inventory directive in kustomization.yaml

New in v2.1.0, a kustomization file may have an inventory field:

inventory:
  type: ConfigMap
  configMap:
    name: prune-cm-name
    namespace: some-namespace

Motivation

If present, kustomize build will make an inventory object, which could be a ConfigMap, or an App (to be added), which can be consumed by a client such as those under development in cli-experimental.

The client can recognize this object by name and use it to do a better job with actions like apply, prune and delete.

Implementation

The inventory ConfigMap contains two special annotations:

  • kustomize.config.k8s.io/Inventory The value of this annotation is the JSON blob for an Inventory object. The Inventory is a struct that contains following information

    • all objects within this kustomization target
    • all objects that reference within this kustomization target

    Here is an example of an Inventory object

    {
  "current":
    {
      "apps_v1_Deployment|default|mysql":null,
      "~G_v1_Secret|default|pass-dfg7h97cf6":
        [
          {
            "group":"apps",
            "version":"v1",
            "kind":"Deployment",
            "name":"mysql",
            "namespace":"default"
          }
        ],
      "~G_v1_Service|default|mysql":null
    }
  }

  • kustomize.config.k8s.io/InventoryHash The value of this annotation is a hash that is computed from the list of items in the Inventory

Basically, this inventory object acts a record of objects that are applied as a group. This object can be consumed by a client such as cli-experimental. The client can recognize the inventory annotations and take proper actions when running apply, prune and delete.

Example

Take following kustomization.yaml as an example

resources:
- deployment.yaml
- service.yaml


secretGenerator:
- name: pass
  literals:
  - password=secret

inventory:
  type: ConfigMap
  configMap:
    name: root-cm
    namespace: default

namespace: default

where the deployment.yaml is

apiVersion: apps/v1
kind: Deployment
metadata:
  name: mysql
  labels:
    app: mysql
spec:
  revisionHistoryLimit: 2
  selector:
    matchLabels:
      app: mysql
  strategy:
    type: Recreate
  template:
    metadata:
      labels:
        app: mysql
    spec:
      containers:
      - image: mysql:5.6
        name: mysql
        env:
        - name: MYSQL_ROOT_PASSWORD
          valueFrom:
            secretKeyRef:
              name: pass
              key: password
        ports:
        - containerPort: 3306
          name: mysql
        volumeMounts:
        - name: mysql-persistent-storage
          mountPath: /var/lib/mysql
      volumes:
      - name: mysql-persistent-storage
        emptyDir: {}

and the service.yaml is

apiVersion: v1
kind: Service
metadata:
  name: mysql
  labels:
    app: mysql
spec:
  ports:
    - port: 3306
  selector:
    app: mysql

Running kustomize build gives 4 objects. Besides the Deployment mysql, the Service mysql, and the Secret pass, the output also contains a ConfigMap object as

apiVersion: v1
kind: ConfigMap
metadata:
  annotations:
    kustomize.config.k8s.io/Inventory: '{"current":{"apps_v1_Deployment|default|mysql":null,"~G_v1_Secret|default|pass-dfg7h97cf6":[{"group":"apps","version":"v1","kind":"Deployment","name":"mysql","namespace":"default"}],"~G_v1_Service|default|mysql":null}}'
    kustomize.config.k8s.io/InventoryHash: 7mgt867b75
  name: haha
  namespace: default

It is clear that this ConfigMap contains an Inventory annotation.

Hash

Note that in the ConfigMap generated from inventory field, there is a hash b965tb9c7d. It is the value for annotation kustomize.config.k8s.io/InventoryHash.

This hash is computed by hashing all the keys in data field, which is the following list in this example.

apps_v1_Deployment|default|mysql
~G_v1_Secret|default|pass-dfg7h97cf6
~G_v1_Service|default|mysql

When any object is added or removed from the kustomzation target, the hash changes. Thus by simply comparing the hash in the inventory objects, one can determine if the list of objects has changed.

How prune works

In cli-experimental, there are different subcommands, apply and prune. Both are able to recognize an inventory object and looking for its existing object on the cluster.

the apply command recognizes the inventory object by the annotation kustomize.config.k8s.io/InventoryHash. It then compares the current hash with the hash for the same object in the cluster. Since the hash reflects if there is any object added or removed, apply takes different actions correspondingly.

  • When there is no existing inventory object in the cluster, apply creates the inventory object.
  • When the current hash is the same as the one in cluster, apply doesn't change the existing object in the cluster.
  • when the current hash is different, apply merges the inventory annotation of the existing object in the cluster and the incoming object. The hash is updated to the latest hash.

The prune command parses the value of kustomize.config.k8s.io/Inventory of the existing inventory object and computes two sets of objects based on the parsed data. To be simple,

  • The items in Inventory.Current will be kept
  • The items in Inventory.Previous will be pruned when they are not needed.