[question]reservation对象和namespace配额疑问

[zj619]

What happened: 在namespace huang中设置限额为request和limit都为1核1Gi，然后创建3个资源预留对象，每个资源预留对象request和limit都为500m 和800Mi，申请出来3个reservation对象都是RUNNING，结果符合预期吗？ 问题1、reservation对象变成avaliable就是占住了资源吗？ 问题2、资源预留对象和namespace配额有无关系？

------------------------------------quota.yaml-------------------------------------------------------------- apiVersion: v1 kind: ResourceQuota metadata: name: huang-quota namespace: huang spec: hard: requests.cpu: "1" requests.memory: "1Gi" limits.cpu: "1" limits.memory: "1Gi"

--------------------------------------------reservation-demo.yaml---------------------------- apiVersion: scheduling.koordinator.sh/v1alpha1 kind: Reservation metadata: name: reservation-huang spec: allocateOnce: false template: # set resource requirements namespace: huang spec: containers:

• name: stress image: polinux/stress imagePullPolicy: IfNotPresent resources: # reserve 500m cpu and 800Mi memory requests: cpu: 500m memory: 800Mi limits: cpu: 500m memory: 800Mi schedulerName: koord-scheduler # use koord-scheduler owners: # set the owner specifications
  • object: # owner pods whose name is default/pod-demo-0 name: pod-huang-0 namespace: huang
  • object: # owner pods whose name is default/pod-demo-0 name: pod-huang-1 namespace: huang ttl: 1h # set the TTL, the reservation will get expired 1 hour late

--------------------------------------------reservation-demo2.yaml---------------------------------------- apiVersion: scheduling.koordinator.sh/v1alpha1 kind: Reservation metadata: name: reservation-huang spec: allocateOnce: false template: # set resource requirements namespace: huang spec: containers:

• name: stress image: polinux/stress imagePullPolicy: IfNotPresent resources: # reserve 500m cpu and 800Mi memory requests: cpu: 500m memory: 800Mi limits: cpu: 500m memory: 800Mi schedulerName: koord-scheduler # use koord-scheduler owners: # set the owner specifications
  • object: # owner pods whose name is default/pod-demo-0 name: pod-huang-0 namespace: huang
  • object: # owner pods whose name is default/pod-demo-0 name: pod-huang-1 namespace: huang ttl: 1h # set the TTL, the reservation will get expired 1 hour late [root@kde-offline3 huang]# ls quota.yaml reservation-demo2.yaml reservation-demo3.yaml reservation-demo.yaml [root@kde-offline3 huang]# cat reservation-demo2.yaml apiVersion: scheduling.koordinator.sh/v1alpha1 kind: Reservation metadata: name: reservation-huang2 spec: allocateOnce: false template: # set resource requirements namespace: huang spec: containers:
• name: stress image: polinux/stress imagePullPolicy: IfNotPresent resources: # reserve 500m cpu and 800Mi memory requests: cpu: 500m memory: 800Mi limits: cpu: 500m memory: 800Mi schedulerName: koord-scheduler # use koord-scheduler owners: # set the owner specifications
  • object: # owner pods whose name is default/pod-demo-0 name: pod-huang2-0 namespace: huang
  • object: # owner pods whose name is default/pod-demo-0 name: pod-huang2-1 namespace: huang ttl: 1h # set the TTL, the reservation will get expired 1 hour later

-------------------------------------------------reservation-demo3.yaml---------------------------------------------------- apiVersion: scheduling.koordinator.sh/v1alpha1 kind: Reservation metadata: name: reservation-huang2 spec: allocateOnce: false template: # set resource requirements namespace: huang spec: containers:

• name: stress image: polinux/stress imagePullPolicy: IfNotPresent resources: # reserve 500m cpu and 800Mi memory requests: cpu: 500m memory: 800Mi limits: cpu: 500m memory: 800Mi schedulerName: koord-scheduler # use koord-scheduler owners: # set the owner specifications
  • object: # owner pods whose name is default/pod-demo-0 name: pod-huang2-0 namespace: huang
  • object: # owner pods whose name is default/pod-demo-0 name: pod-huang2-1 namespace: huang ttl: 1h # set the TTL, the reservation will get expired 1 hour later

[root@kde-offline3 huang]# ls quota.yaml reservation-demo2.yaml reservation-demo3.yaml reservation-demo.yaml [root@kde-offline3 huang]# cat reservation-demo3.yaml apiVersion: scheduling.koordinator.sh/v1alpha1 kind: Reservation metadata: name: reservation-huang3 spec: allocateOnce: false template: # set resource requirements namespace: huang spec: containers:

• name: stress image: polinux/stress imagePullPolicy: IfNotPresent resources: # reserve 500m cpu and 800Mi memory requests: cpu: 500m memory: 800Mi limits: cpu: 500m memory: 800Mi schedulerName: koord-scheduler # use koord-scheduler owners: # set the owner specifications
  • object: # owner pods whose name is default/pod-demo-0 name: pod-huang3-0 namespace: huang
  • object: # owner pods whose name is default/pod-demo-0 name: pod-huang3-1 namespace: huang ttl: 1h # set the TTL, the reservation will get expired 1 hour later

What you expected to happen:

Environment:

• Koordinator version: - v1.4.0
• Kubernetes version (use kubectl version): v1.22.3
• docker/containerd version: containerd 1.5.0
• OS (e.g: cat /etc/os-release): Ubuntu 20.04.4 LTS
• Kernel (e.g. uname -a): Linux 5.10.112-11.al8.x86_64 #1 SMP Tue May 24 16:05:50 CST 2022 x86_64 x86_64 x86_64 GNU/Linux

Anything else we need to know:

[saintube]

@zj619 Reservation 对象是 non-namespaced，不过它在调度器内的 fake pod 是依据 template 生成的，可以填写 namespace，从而生效一些 namespace 维度的调度功能。针对你的问题，这里有2点需要注意下：

1. reservation.spec.template 中的 namespace 填写有误，namespace 字段需要填写在 metadata 字段下。因为 reservation CRD 设置了不会强校验字段合法，因此你这里错误的 namespace 最终也可以提交成功，但实际生产的 fake pod 不会生效指定的 namespace。
2. reservation 通过生成 fake pod 在调度流程里预留资源，但 fake pod 对外部组件是不可见的，所以 webhook 等流程默认不会处理 reservation。