on 19-Jun-2019 (Wed)

Question

In kubernetes, [...] can be thought of as a logical server that contain one or more (but mostly one) running applications, running as a container.

Answer

pods

Question

In kubernetes, [...] are actual (physical or cloud) servers that run one or more pods.

Answer

nodes

Question

In kubernetes, [...] are wrappers around a container and can be though of as a logical server, while nodes are the actual phyisical or cloud servers that run them.

Answer

pods

Question

Answer

led to

Question

In kubernetes, if you want to create a deployment of pods, the best/easiest/fastest command to use is: "kubectl [...] " command

Answer

run

^^^ e.g.: kubectl run kubernetes-bootcamp --image=gcr.io/google-samples/kubernetes-bootcamp:v1 --port=8080 --replicas=3

Question

In kubernetes, a set of pods all running the same container app, is called a [...] of pods.

Answer

replica

Question

In kubernetes, if you have a set of pods (each with a container running the same app), you can group them together for the purpose of internal, or more importantly, external exposure via a kubernetes concept called [...]

Answer

service

^^ note that service is created behind the scenes via issueing the "kubectl expose" command.

Question

In kubernetes, a [...] is a way to logically group a set of pods together for the purpose of allowing them to be accessed internally (within the cluster) and more importantly externally (from outside the cluster).

Answer

service

Question

In kubernetes, issue cammand to run a three pod deployment of the following docker image: "gcr.io/google-samples/kubernetes-bootcamp:v1". You can use port "8080" and name the deployment "kubernetes-bootcamp"

Answer

kubectl run kubernetes-bootcamp --image=gcr.io/google-samples/kubernetes-bootcamp:v1 --port=8080 --replicas=3

Question

In kubernetes, while the "kubectl run" command will create a deployment based on a docker image provided (e.g. : kubectl run kubernetes-bootcamp --image=gcr.io/google-samples/kubernetes-bootcamp:v1 --port=8080 --replicas=3), the kubectl [...] command will create a service from these 3 pods so they can be accessed from outside the cluster

Answer

expose

Question

In kubernetes, let say you do a 3 pod deployment called "kubernetes-bootcamp" using docker image "gcr.io/google-samples/kubernetes-bootcamp:v1" as follows:

kubectl run kubernetes-bootcamp --image=gcr.io/google-samples/kubernetes-bootcamp:v1 --port=8080 --replicas=3

Run command to get these pods to be accessible from a dynamic port on the host machine:

Answer

kubectl expose deployment kubernetes-bootcamp --type=NodePort --port=8080

^^ the key point here is knowing the "kubectl expose" command and how to use it

^^^ note above command creates a service (also called "kubernetes-bootcamp"), which allows the pods to be exposed via a dynamic port, e.g. via: "curl $(hostname):30993", where 30993 is the dynamic port generated.

^^^^ note the --type param is set to NodePort, which will expose the nodes via a dynamiclly generated port, so you can access from the node/host you are running kubectl from. There are other more advanced/better --type options, refer to the k8s docs

Question

In kubernets, [...] s are used to add metadata key/value pairs to resrouces (such as pods), which can then be used for easier searching/grouping of these resources.

Answer

labels

Question

In kubernetes, say you have a pod called "kevintest-5sdflkj-923s" (just FYI: "kevintest" is the associated deployment name), issue command to add a app=v1 metadata tag to this pod:

Answer

kubectl label pod kevintest-5sdflkj-923s app=v1

^^ note the format follows the "kubectl action resrouce" formula, where "label" is action, "pod" is resource, then comes the pod name and the label key/value to add

^^^ note that ofcourse labelling can also be done in the deployment .yaml file, if you use this approach for deployment.

Question

In kubernetes, issue command to get simple list of all pods that have the following label (i.e.metadata tag):
app=v1

Answer

kubectl get pods -l app=v1

Question

In kubernetes, scaling (up or down) an app is accomplished by changing the number of [...] s in a Deployment. Hint: answer is not "pod", as that is too obvious.

Answer

replicas

Question

In kubernetes, when you use the "kubectl scale" command to scale up/down your pods, this is reflected back when you use the "kubectl get [...] (s)" command

Answer

deployment(s)

Question

In kubernetes, when you scale up your deployment to use multiple pods, load-balancing is automatically handled by the kubernetes [...] used to expose the deployment (to external traffic).

Answer

service

Question

In kuberentes, let say you setup a one pod deployment with the following command:

kubectl run kubernetes-bootcamp --image=gcr.io/google-samples/kubernetes-bootcamp:v1 --port=8080

Issue command to scale up the deployment to 4 pods

Answer

kubectl scale deployment kubernetes-bootcamp --replicas=4

^^ after running above command, when you then run "kubectl get deployment kubernetes-bootcamp", it will show that 4 pods are in READY state in the deployment now

Question

In kubernets, explain in detail what happens, by default, when you do a rolling update via the "kubectl set image" command?

Answer

One old pod is taken down and a whole new pod is created in its place with the update image, this continues until all old pods are replaced with new pods with the new image. By default 1 pod is taken down at a time at most, and 1 new one created at most at a time, but this is configurable.

Question

In kubernetes, you can initiate a rolling update (where, by default, one pod is taken down and new one created with new conatiner image), by using the "kubectl [...] [...] " command

Answer

set image

^^ e.g.

kubectl set image deployment.v1.apps/kubernetes-bootcamp kubernetes-bootcamp=gcr.io/google-sample/kubernetes-bootcamp:v2

, where kubernetes-bootcamp= referes to container image name which you can get from "kubectl describe deploymen kubernetes-bootcamp"

Question

In kubernetes, if you do a rolling update (via "kubectl set image" command) and things go bad (i.e. one or more pods are no longer available), you can rollback via the "kubectl [...] [...] " command

Answer

kubectl rollout undo

^^ you can either rollback to immediate previous version, e.g.: kubectl rollout undo deployment kubernetes-bootcamp

^^^ or you can rollback to a specific previous version number, e.g.: kubectl rollout undo deployment kubernetes-bootcamp --to-revision=2

^^^^ to get the previuos versions/revision of a deployment, you can do "kubectl rollout history deployment kubernetes-bootcamp"

Question

- In kubernetes, if you do several rolling updates (via "kubectl set image" commands), you can check the deployment version history via the "kubectl [...] [...] " command.

Answer

kubectl rollout history

^^ e.g.: kubectl rollout history deployment kubernetes-bootcamp

Question

In kubernetes, let's say you have 4 pods, in deployment "kubernetes-bootcamp", that are running container image "gcr.io/google-sample/kubernetes-bootcamp:v1", issue command to do rolling upgrade to image "gcr.io/google-sample/kubernetes-bootcamp:v2", fillout below occultions:

kubectl [...] [...] deployment.v1.apps/kubernetes-bootcamp kubernetes-bootcamp= [...]

NOTE: the kubernetes-bootcamp= part is referring to the conatiner name (which you can get from "kubectl describe deployment kubernetes-bootcamp"

Answer

kubectl set image deployment.v1.apps/kubernetes-bootcamp kubernetes-bootcamp=gcr.io/google-sample/kubernetes-bootcamp:v2

Question

In kubernetes, if you do a rolling update to upgrade container images of your pods via following command (note deployment is called "kubernetes-bootcamp"):

kubectl set image deployment.v1.apps/kubernetes-bootcamp kubernetes-bootcamp=gcr.io/google-sample/kubernetes-bootcamp:v2

Assume update went bad (i.e. one or more pods are down post-update), issue command to roll back to imediate previous version

Answer

kubectl rollout undo deployment kubernetes-bootcamp