CODEWITHSUNDEEP
kubernetes
user642318
user642318

Reputation: 417

How do you specify cpu or memory resource on container in kurbenetes

The definition of the pod which is taken from https://github.com/mesosphere/kubernetes-mesos/blob/master/examples/pod-nginx.json

{  
   "kind":"Pod",
   "apiVersion":"v1beta1",
   "id":"nginx-id-01",
   "desiredState":{  
      "manifest":{  
         "version":"v1beta1",
         "containers":[  
            {  
               "name":"nginx-01",
               "image":"dockerfile/nginx",
               "ports":[  
                  {  
                     "containerPort":80,
                     "hostPort":31000
                  }
               ],
               "livenessProbe":{  
                  "enabled":true,
                  "type":"http",
                  "initialDelaySeconds":30,
                  "httpGet":{  
                     "path":"/index.html",
                     "port":"8081"
                  }
               }
            }
         ]
      }
   },
   "labels":{  
      "name":"foo",
      "cluster":"gce"
   }
}

I cant seem to find how can you specify resources.

Upvotes: 0

Views: 1439

Answers (4)

Shifali Srivastava
Shifali Srivastava

Reputation: 99

You can specify the cpu or resources for any container in the deployment file itself in your container specification .

nano task2deploy1.yaml

apiVersion: extensions/v1beta1
kind: Deployment
metadata:
  name: task2deploy1
spec:
  replicas: 1
  template:
    metadata:
     labels:
       app: task2deploy1
     spec:
  containers:
  - name: task2deploy1
    image: nginx
    ports:
    - containerPort: 80
    resources:
      requests:
        memory: "64Mi"
        cpu: "250m"
      limits:
        memory: "128Mi"
        cpu: "500m"

This means the request of 64Mi memory and 250m cpu can be sent to container named task2deploy1.

The container although has a limit of 128Mi memory and 500m cpu.

 kubectl create –f task2deploy1.yaml –namespace=Shifali

Once you create pods with the help of this deployment the pod will have these resources as per mentioned in the deployment file .

 kubectl describe pod task2deploy1-765cfc9b89-xrgsr --namespace=Shifali

Name:           task2deploy1-765cfc9b89-xrgsr

Limits:

  cpu:     500m
  memory:  128Mi

Requests:
  cpu:        250m
  memory:     64Mi

along with this more information would be displayed but what need to see is the resources.

Upvotes: 0

Margach Chris
Margach Chris

Reputation: 1729

Officially, you can add a resource field in each container definition in your pod, you can add requests and limits:

resources:
  requests:
    memory: "qMi"
    cpu: "pm"
  limits:
    memory: "yMi"
    cpu: "x"

The sum of these values is then summed up with other containers in the pod to calculate the total request and limit per pod.

Please take a look at https://kubernetes.io/docs/concepts/configuration/manage-compute-resources-container/

Upvotes: 0

lavalamp
lavalamp

Reputation: 399

You can enter values for memory & CPU in your container spec: https://github.com/GoogleCloudPlatform/kubernetes/blob/master/pkg/api/v1beta1/types.go#L257

Our scheduler doesn't do much with them yet, but it also doesn't completely ignore them any more.

Upvotes: 1

Adam
Adam

Reputation: 4322

You can see examples of Kubernetes resources being specified at https://github.com/GoogleCloudPlatform/kubernetes/blob/master/examples/guestbook/frontend-controller.json#L16

However, Kubernetes (hence Kubernetes-Mesos) doesn't really do anything with these resources yet, as per https://github.com/GoogleCloudPlatform/kubernetes/issues/168

Once this is implemented in Kubernetes, the Kubernetes-Mesos framework will support it as well: https://github.com/mesosphere/kubernetes-mesos/issues/76

Upvotes: 1

Related Questions