Config - Resource Requirements

Question 1

What happens to the resources on a node, when a pod gets deployed on it?

Answer 1

• ## the pod consumes part of the resources that are available to the node

Question 2

Who decides where pods are deployed? On what is part of this consideration?

Answer 2

• kube-scheduler decides this
• considers the current resource utilization of the node
• identifies the best node for pod deployment

Question 3

What happens if nodes have not sufficient resources available?

Answer 3

• kube-scheduler avoids these nodes
• uses a node with sufficient resources
• if there are no nodes with sufficient resources available, it holds back the pod (Status: Pending)

Question 4

What is the amount of needed CPU/Memory/… called for a pod?

Answer 4

Resource Request

Question 5

How do we add a Resource Request to a pod?

Answer 5

pod-yaml
~~~
apiVersion: v1
kind: Pod
metadata:
name:
labels:
spec:
containers:
- name:
image:
ports:
resources:
requests:
memory: “4Gi”
cpu: 2
~~~

Question 6

What are valid cpu values for resource requests?

Answer 6

0.1 - full int value
0.1 equals 100m
lowest value is 1m
1 count of cpu is equivalent of vCPU (AWS, CGP, Azure, …)

Question 7

What are valid memory values for resource requests?

Answer 7

256 Mi
268M
1G

Question 8

When specifying resouce requests for memory, what is the difference between G/M/K and Gi/Mi/Ki

Answer 8

G= Gigabytes = 1,000,000,000 bytes

Gi = Gigibytes = 1,074,741,824 bytes

Question 9

What are the regular limitations for a node what resources it can consume on a node?

Answer 9

By default no limit

Question 10

How can we limit, what a pod should be allowed to consume on a node?

Answer 10

by setting/using Resource Limits
in pod-yaml, under resources
~~~
spec:
resources:
requests:
cpu: 2
memory: “1Gi”
limits:
cpu: 3
memory: “2Gi”
~~~

Question 11

What happens, when a pod tries to exceed the set limit of resource consumption?

Answer 11

• cpu: consumption gets throttled not to exceed the limit
• memory: container can consume more memory than its limit
• if this happens constantly, the pod gets terminated with Out Of Memory Error (OOM)

Question 12

What is an implication of the default Kubernetes behavior looking at resource limits and requests

Answer 12

• by default no limits or requests
• pods can suffocate each other by rising resource utiliation

Question 13

What happens if we only set limits for a pod, but no resource requests?

Answer 13

• limits are assumed to be also the request value

Question 14

How do we let one pod consume cpu resources, higher than its limit, as long as the other pod on the node are not using all thats left?

Answer 14

Only setting requests, no limits
Usually most ideal set-up

Question 15

How do we ensure that every pod has some limit set?

Answer 15

By using LimitRanges
Help you define default values to be set for pods.
Set at namespace level
Is an object, created by yaml

Question 16

How does a Limit Range yaml look like?

Answer 16

apiVersion: v1
kind: LimitRange
metadata:
 name: cpu-resource-constraint
spec:
 limits:
  - default:
      cpu: 500m
		defaultRequest:
		  cpu: 500m
		max:
		  cpu: "1"
		min: 
		  cpu: 100m
		type: container

Default and max refer to the limit

Question 17

When are the limits, set by a limit range, enforced?

Answer 17

When a pod gets created

Question 18

How can we limit the total amount of resources that can be consumed by applications deployed in a Kubernetes cluster?

Answer 18

yes: Resource Quota, at namespace level

I.E. all pods together should consume max this amount of memory

Question 19

What is a resource quota and how does it look?

Answer 19

• namespace level object
• limits the total amount of resources used by all pods
  own yaml:

apiVersion: v1
kind: ResourceQuota
metadata: 
 name: my-resource-quota
spec:
  hard:
	  requests.cpu: 4
		requests.memory: 4Gi
		limits.cpu: 10
		limits:memory: 10Gi