Profile-Based Limits

02-profiles removes the spec.limits block entirely. The squad picks a tier name and Orkestra expands the ResourceQuota at reconcile time.

apiVersion: provisioner.orkestra.io/v1
kind: NamespaceClaim
metadata:
  name: team-beta-claim
spec:
  targetNamespace: team-beta
  team: beta
  owner: beta-operator
  ownerNamespace: default
  tier: medium

Four fields. No quota knowledge required from the CR author.

What changes in the Katalog

The only difference from 01-explicit is the ResourceQuota declaration:

# 01-explicit
resourceQuotas:
  - name: "{{ .spec.team }}-quota"
    hard:
      pods: "{{ .spec.limits.maxPods }}"
      requests.cpu: "{{ .spec.limits.cpu }}"
      requests.memory: "{{ .spec.limits.memory }}"

# 02-profiles
resourceQuotas:
  - name: "{{ .spec.team }}-quota"
    profile: "{{ .spec.tier }}"

Orkestra expands medium to pods: 20, requests.cpu: 4, requests.memory: 8Gi before the ResourceQuota is applied. ork validate rejects unknown tier names at load time.

Tier	Pods	CPU	Memory
`small`	10	2	4Gi
`medium`	20	4	8Gi
`large`	50	8	16Gi
`xlarge`	100	16	32Gi

Try it

cd 02-profiles
ork validate
ork simulate
ork run

After ork run, confirm the profile expanded correctly:

kubectl get resourcequota beta-quota -n team-beta -o jsonpath='{.spec.hard.pods}'
# 20
kubectl get resourcequota beta-quota -n team-beta -o jsonpath='{.spec.hard.requests\.cpu}'
# 4

Run both together

From the root of the example:

cd ..
ork simulate   # runs both sub-examples
ork e2e        # provisions two kind clusters in sequence