Skip to main content

스케쥴링

· 4 min read

Labels and Selectors​

We have deployed a number of PODs. They are labelled with tier, env and bu. How many PODs exist in the dev environment (env)?

kubectl get pods --selector env=dev

경험에 따르면, CPU sort 하는 것도 시험에 나온다고 하였다.

kubectl top po -A --sort-by=cpu

How many PODs are in the finance business unit (bu)?

kubectl get pods --selector bu=finance

How many objects are in the prod environment including PODs, ReplicaSets and any other objects?

kubectl get all --selector env=prod

Identify the POD which is part of the prod environment, the finance BU and of frontend tier?

kubectl get all --selector env=prod,bu=finance,tier=frontend

replicaset-definition.yaml

apiVersion : apps/v1
kind : ReplicaSet
metadata:
name: simple-webapp
labels:
app: app1
function: Front-end
spec:
replicas: 3
selector:
matchLabels:
app: App1
template:
metadata:
labels:
app: App1
function: Front-end
spec:
containers:
- name : simple-webapp
image: simple-webapp

service-definition.yaml

apiVersion : v1
kind : Service
metadata:
name: my-service
spec:
selector:
app: App1
ports:
- protocol:TCP
port : 80
targetPort: 9376

Node Selector​

kubectl label nodes node-name label-key=label-value

ex) kubectl label nodes node-1 size=Large 라고 설정을 하고

spec:
containers:
- name : data-processor
image: data-processor
nodeSelector: Large

Node Affinitiy​

Node Affinity Type​

requiredDuringSchedulingIngnoredDuringExecution - Type1 : 강제성이 있음.
preferredDuringSchedulingIngnoredDuringExecution - Type2 : 어쩔 수 없으면 함.

Apply a label color=blue to node node01

kubectl label nodes node01 color=blue

Create a new deployment named blue with the nginx image and 3 replicas.

kubectl create deployment blue --image=nginx
kubectl scale deployment/blue --replicas=3

Which nodes can the pods for the blue deployment be placed on?

Make sure to check taints on both nodes!

kubectl describe node controlplane | grep Taints
kubectl describe node node1 | grep Tains

check Taint both!

Create a new deployment named red with the nginx image and 2 replicas, and ensure it gets placed on the controlplane node only.

Use the label key - node-role.kubernetes.io/control-plane - which is already set on the controlplane node.

kubectl create deployment red --image=nginx --replicas=2 --dry-run=client -o yaml  

kubectl create deployment red --image=nginx --replicas=2 --dry-run=client -o yaml > red.yaml


affinity:
nodeAffinity:
requiredDuringSchedulingIgnoredDuringExecution:
nodeSelectorTerms:
- matchExpressions:
- key: disktype
operator: In
values:
- ssd

Resource Requirements and Limit​

개념) Kube-Scheduler가 노드의 Mem / CPU를 보고 pod를 Node에 할당

할당 단위​

CPU : 1m(최소 단위 이 밑으로 내려갈 수 없음.) 1 (CPU Core를 의미)

The elephant pod runs a process that consumes 15Mi of memory. Increase the limit of the elephant pod to 20Mi.

Delete and recreate the pod if required. Do not modify anything other than the required fields.

kubectl create pod elephant --image=polinux/stress  --dry-run=client -o yaml > elephant.yaml

kubectl replace -f elephant.yaml --force

Daemon Sets​

모든 노드에 각각 존재함. 모니터링에서 사용됨.

kube-scheduler의 영향을 받지 않음

 kubectl get daemonsets --all-namespaces

Which namespace is the kube-proxy Daemonset created in?

What is the image used by the POD deployed by the kube-flannel-ds DaemonSet?

kubectl describe daemonset kube-flannel-ds --namespace=kube-flannel

Deploy a DaemonSet for FluentD Logging.

Name: elasticsearch

Namespace: kube-system

Image: registry.k8s.io/fluentd-elasticsearch:1.20 kubectl create Daemonset elasticsearch --namespace=kube-system --image=registry.k8s.io/fluentd-elasticsearch:1.20 -o yaml > elasticsearch.yaml

kubectl create Daemonset은 없음

kubectl create deployment elasticsearch --namespace=kube-system --image=registry.k8s.io/fluentd-elasticsearch:1.20 --dry-run=client -o yaml

kubectl create deployment elasticsearch --namespace=kube-system --image=registry.k8s.io/fluentd-elasticsearch:1.20 --dry-run=client -o yaml > fluentD.yaml

Static Pod​

각 node 안 kubelet에서 생성을 함.

kube-scheduler의 영향을 받지 않음

pod.yaml은 etc/kubernetes/manifests 에 존재하며 이를 가져온다.

Multiple Scheduler​

scheduler 만들기

apiVersion: kubescheduler.config.k8s.io/v1
kind: KubeSchedulerConfiguration
profiles:
- schedulerName: my-scheduler
leaderElection:
leaderElect: false

Pod에 custom scheduler 적용하는 방법

apiVersion : v1
kind : Pod
metadata:
name: nginx
spec:
containers:
- image: nginx
name: nginx
shcedulerName: my-custom-scheduler

View Events​

kubectl get evnets -o wide

kubectl logs my-custom-scheulder --name-space=kube-system

What is the name of the POD that deploys the default kubernetes scheduler in this environment?

kubectl get pods --namespace=kube-system

kubectl get pods -A

Deploy an additional scheduler to the cluster following the given specification.

Use the manifest file provided at /root/my-scheduler.yaml. Use the same image as used by the default kubernetes scheduler.

Scheduling 4단계​

Scheduling Queue → Filtering → Scoring → Binding

Scheduling Queue : PrioritySort

Filtering : NodeResourcesFit, NodeName, NodeUnshcedulable

Scoring : NodeResourcesFit, ImageLocality

Binding : DefaultBinder