Basic Concepts

Question 1

What are the elements of the control plane?

Answer 1

• etcd (KV store)
• API server
• scheduler
• controller manager
• cloud controller manager

Question 2

etcd

Answer 2

• distributed KV store for cluster state
• how many resources are running, status of resources, etc
• stores and replicates all cluster states
• “et-see-dee”
• distributed etc directory

Question 3

API server

Answer 3

used by pods and users to communicate with the control plane
accepts REST commands to interact with cluster resources

Question 4

Controller Manager

Answer 4

• control loop that makes sure the etcd state matches the node state
• for example, if a workflow requires 3 replicas and 1 replica is down, the controller manager sends a command to restore it

Question 5

Worker Nodes

Answer 5

• the compute on which containers are executed
• can be a physical or virtual machine depending on the cluster
• managed by the control plane and contain the necessary services to run pods
• if nodes are unavailable, kubernetes cannot run workloads

Question 6

What is kubernetes?

Answer 6

• An orchestration platform for running containers
• Runs workloads by placing containers into pods to run on nodes
• Schedules pods on nodes on a best-effort basis

Question 7

What are the elements of a node?

Answer 7

All nodes need to be bootstrapped with
- kubeproxy
- kubelet
- container runtime (e.g. docker)

Question 8

Cardoning

Answer 8

Stop new workflows from being scheduled

Question 9

Draining

Answer 9

Start moving workflows off a given node

Question 10

What is a kubernetes cluster?

Answer 10

• Physical isolation of hardware (set of connected nodes)
• Combination of control plane and worker nodes

Question 11

What is the control plane?

Answer 11

A controller that runs in a loop trying to ensure that the cluster is in a desired state

Question 12

Namespaces

Answer 12

• Provide logical isolation of workloads
• Helpful for provisioning cluster resources
• Kubernetes resources must live within a namespace unless they are cluster-scoped

Question 13

What is tainting?

Answer 13

Allows for customizable scheduling of nodes
When you taint a node, you need a pod to tolerate that taint
Taints and tolerations are how pods are scheduled on nodes

Question 14

kubeproxy

Answer 14

• responsible for communication between the control plane and the node itself
• network proxy that runs on each node in a cluster o maintain network rules and enable network communication

Question 15

kubelet

Answer 15

• takes instructions from control plane and executes them on the node
• agent that gets the pod config from the api server and ensures the pods and container state have the desired config

Question 16

kubectl

Answer 16

• kube CTL
• application to interact with the kubernetes cluster

Question 17

Pods

Answer 17

• Smallest deployable units of compute that you can deploy and manage in Kubernetes
• Run the workflow
• Group of containers with shared storage and network resources and a specification for how to run the containers

Question 18

When is a pod considered up?

Answer 18

When all the containers are in a running state

Question 19

What are initcontainers?

Answer 19

• Allow preconfigurations such as bootstrap and auth credentials

Question 20

What are affinity and non-affinity?

Answer 20

How pods are scheduled with respect to other pods

Question 21

What is a readiness probe?

Answer 21

Used to determine whether a pod is ready to accept traffic

Question 22

What is a liveness probe?

Answer 22

Used to determine whether or not a pod is healthy

Question 23

kube scheduler

Answer 23

• background process that watches for new kubernetes pods with no assigned nodes and assigns them to a worker node for execution

Question 24

controller

Answer 24

manage kubernetes jobs

Question 25

helm

Answer 25

- de facto tool for deploying application stacks to kubernetes - templating engine and package manager for a set of Kubernetes manifests - at runtime, it replaces placeholders in YAML template files with actual user-defined values

Question 26

container runtime engine

Answer 26

- software responsible for managing containers (e.g. Docker)

Question 27

chart file

Answer 27

- contains manifest files, default configuration values, and metadata - artifact produced by helm executable - bundles the manifests that comprise the API resources of an application

Question 28

kubernetes operator

Answer 28

software extensions to kubernetes that use custome resources to manage applications and their components

Question 29

Pod pending

Answer 29

The pod has been accepted by the Kubernetes system but one or more of the container images has not been created

Question 30

Pod running

Answer 30

At least one container is still running or is in the process of starting or restarting

Question 31

Pod succeeded

Answer 31

All containers in the pod terminated successfully

Question 32

Pod failed

Answer 32

Containers in the pod terminated; at least one failed with an error

Question 33

Pod unknown

Answer 33

The pod state cannot be obtained

Question 34

kubectl logs -p

Answer 34

Show logs from previous container

Question 35

kubectl exec -it hazelcast -- /bin/sh

Answer 35

exec: open a shell it: interactive; remove it to only pass in one command hazelcast: name of the pod --: separate the exec command options from the command you want to run inside the container /bin/sh: command to run inside the container

Question 36

Are pod ip addresses stable?

Answer 36

No. A pod restart leases a new IP address. That's why services exist. To provide a stable network interface.