(azure) kubernetes (service)

Question 1

what is the difference between a deployment and service?

Answer 1

a deployment is used to keep a set of pods running by creating pods from a template a service is used to allow network access to a set of pods

a service is an abstract way to expose an application running on a set of Pods as a network service.

Question 2

what do targetport and port enable in a service?

kind: Service
metadata:
name: vanilla-webapp
spec:
type: LoadBalancer
ports:
- port: 80
targetPort: 5000

Answer 2

targetport

port to listen to (internally) against any container with label ‘vanilla-webapp’

port

port to listen to and expose to outside world

Question 3

what is a container orchestrator?

Answer 3

centralised management layer for containerised applications

Question 4

what is a cluster?

Answer 4

• a cluster is a group of compute/nodes working together as a single system
• shared networking
• centralised management

Question 5

what is in the control plane?

Answer 5

kube-apiserver

• communication hub in and out of cluster
• exposes k8s API
• cluster gateway
• gatekeeper for authentication
• user entry point for user requests (ui, api, cli)

e.g. user request -> apiserver validates request -> scheduler -> kublet

etcd

• a key-value backing store that holds stated of the entire cluster
• the “cluster brain”
• records all cluster state changes

scheduler

• “intelligent decision-maker”
• determines the best location on the nodes for pods/containers
• affinity and anti-affinity

controller

• monitors and responds to events to maintain desired state
• detects cluster changes
• attempts to recover cluster state to desired state ASAP

e.g. controller -> scheduler -> kublet

Question 6

what is a clusterIP?

Answer 6

internal IP to pod

Question 7

what is a nodePort?

Answer 7

static port to pod

Question 8

what is a loadBalancer?

Answer 8

cloud provided loadBalancer for external access

Question 9

what is an ephemeral volume?

Answer 9

• lasts lifetime of the pod
• will not lose data if pod is destroyed
• eg. caching

Question 10

what is a persistent volume?

Answer 10

• lifetime does not follow lifetime of pod
• will live on after pod destroyed
• eg. databases and messages queues

Question 11

what is a namespace?

Answer 11

• Namespaces help pod-to-pod communication using the same namespace.
• Namespaces are virtual clusters that can sit on top of the same physical cluster.
• They provide logical separation between the teams and their environments.

• Everything in the default or a single namespace would get complex and confusing
• Group resources into namespaces (similar to resource groups)
• Officially: don’t bother if you have a smaller project with less than 10 users
• Unofficially: use namespaces always as a best practice
• Multiple teams is another good use case
• Allows same named applications to co-exist as they are in different namespaces
• Different environments in the same cluster
• This way you can deploy common resources eg: elasticstack, nginx once
• Blue/green deployment
• Access and resource limits on namespaces
• Two teams working on same cluster different namespaces
• Allow access to ONLY their own namespace
• Limit resources on each namespace
• Consider:
  
  • CANT access MOST resources from another namespace
  • Each namespace must define its own configmap, secret
  • CAN share a service across namespaces
  • Some components can NOT be added to a namaespace and must remain global
    
    • Kubectl api-resources –namespaced=false
• Can change active (default) namespace using kubens application

Question 12

what is a configMap used for?

Answer 12

storing non-sensitive configuration data

Question 14

what is a headless service?

Answer 14

a “stateful” service.

think quorum and leaders.

think zookeeper.

a service that has a master and slave nodes

the master must commit before the slaves

this is useful for when you want to scale a resource that is replicated, requires ACID transactions, must remain consistent like a database