My Workplay on docker
-
docker run:- runs a command in a new container .docker run = docker create + docker start -
docker run -p <localhostport>:<containerport> <imagename/id>:- running on ports -
docker ps:- to list all the running containers -
docker ps --all:- list all the container ever created -
docker system prune:- to delete all the containers ever created along with some other properties -
docker logs <container-id>:- to get the logs -
docker start:- start stopped container -
docker stop:- stop the container - gets a sigterm message - terminate signal -
docker kill:- kills the container or stops the container instantly -
docker exec -it <container id> <command>:- Execute an additional command in container.-itmakes us to provide the input.-it equivalent to -i -t -
docker exec -it <container id> sh:- provides access to the terminal inside the context of the container -
docker build .:- build an image from a Docker file -
docker-compose up:- aggregates the output of each container. Similar to docker run myimage -
docker-compose up --build:- similar to docker build and docker run. Rebuilds the container after making any changes to the file -
docker-compose up -d:- starts the containers in the background and leaves them running -
docker-compose down:- stops the running containers at the same time -
docker-compose ps:- show the status of the containers -
docker commit:- manual image generation -
docker build -f <filename> .:- to run a dockerfile with some different name -
docker pull:- pulls an image from registry -
docker push:- pushes an image to registry -
docker search:- search for an image in Docker Hub -
docker history:- shows the history of the image -
docker info:- shows system wide information -
docker rm:- remove one or more containers -
docker rmi:-remove one or more images -
docker pause:- pauses all processes within one or more containers -
docker unpause:- unpause all processes within one or more containers
-
What is Kubernetes - System for running many different containers over multiple different machines
-
Why Kubernetes - When you need to run many different containers with different images
-
Install kubectl - cli to interact with master
-
Install a VM driver virtual box - make VM which will be our single node
-
Install minikube - runs a single node on that vm
-
kubectl:- use for managing containers in the node -
minikube:- use for managing the VM itself (Local Only) -
minikube start:- to start the minikube
| Docker Compose | Kubernetes |
|---|---|
| Each entry can optionally get docker compose to build an image | Kubernetes expects all images to already be built |
| Each entry represents a container we want to create | One config file per object we want to create |
| Each entry defines the n/w requirements (Ports) | We have to manually sets up all n/w |
Get a simple container running on our local K8s cluster running
-
Make sure our image is hosted on docker hub
-
Make one config file to create the container
-
Make one config file to setup n/w
Config file used to create Objects for example StatefulSet, ReplicaController, Pod, Service. Object will serve different functionalities like running container, monitoring a container, setting up nw etc.
-
Pods :- Runs one or more closely related containers
-
Services :- Sets up networking in K8S cluster. There are 4 subtypes
-
- ClusterIp
-
- NodePort : Exposes a container to the outside world and its only good for dev purposes
-
- LoadBalancer
-
- Ingress
-
kubectl apply -f <fileNamePath>:- feed a config file to kubectl -
kubectl get pods:- prints the status of all running pods -
kubectl get services:- prints the status of all running services -
minikube ip:- to get the ip of the VM
| Important Takeaways |
|---|
| Kubernetes is a system to deploy containerized apps |
| Nodes are individual machines or vm's that run containers |
| Masters are machines or vm's with a set of programs to manage nodes |
| Kubernetes didn't build our images - it got them from somewhere else |
| Kubernetes (the master) decided where to run each container - each node can run a dissimilar set of containers |
| To deploy something, we update the desired state of the master with a config file |
| The master works constantly to meet you desired state |
kubectl describe <ObjectType> <ObjectName>:- describes the logs about the specified object type and name
| Limitations on Updating Config file |
|---|
| Can only change spec.containers[*].image |
| Can only change spec.initContainers[*].image |
| Can only change spec.activeDeadlineSeconds |
| Can only change spec.tolerations |
- Solution to this is by using Deployment Object type
| Pods | Deployment |
|---|---|
| Run a single set of containers that are very close | Run a set of identical pods (one or more) |
| Good for one-off dev purposes | Monitors the state of each pod updating as necessary |
| Rarely used directly in production | Good for dev and production |
kubectl delete -f <configfile> :- remove an object (like an imperative update)
kubectl get deployments:- prints the status of the deployment
-
Manually delete pods to get the deployment to recreate them with the latest version but deleting pods manually seems silly(bad idea)
-
Tag images with real version number and specify them in config file but it adds an extra step in the production deployment process(no friendly)
-
Use an imperative command to update the image version the deployment must use also not the best solution we can say
-
kubectl set image <objectType> / <objectName> <containerName> = <newImageToUse>:- Imperative command to update the image
eval $(minikube docker-env)- this only configures your current terminal window ie this is not permanent and have to rerun the same command everytime if we close the terminal window
minikube docker-env
#produces the following output
export DOCKER_TLS_VERIFY="1"
export DOCKER_HOST="tcp://ip:2376"
export DOCKER_CERT_PATH="path"
export MINIKUBE_ACTIVE_DOCKERD="minikube"
# To point your shell to minikube's docker-daemon, run:
# eval $(minikube -p minikube docker-env)
- Use all the same debugging techniques we learned with Docker CLI (Many of these are available through kubectl)
For example kubectl logs or kubectl exec -t sh to enable a shell etc.
-
Manually kill containers to test Kubernetes self heal
-
Delete cached images in the node (
docker system prune -a)
-
Create config file for each service and deployment
-
Test locally on minikube
-
Setup github/travis flow to build images and deploy
-
Deploy app to cloud provider
-
ClusterIp exposes a set of pods to other objects inside the cluster and no outside world
-
NodePort exposes a set of pods to the outside world and good for dev purposes only.
Here we need to give --- 3 consecutive dashes to separate them
apiVersion: apps/v1 kind: Deployment metadata: name: server-deployment spec: replicas: 3 selector: matchLabels: component: server template: metadata: labels: component: server spec: containers: - name: server image: imageName ports: - containerPort: 5000 --- apiVersion: v1 kind: Service metadata: name: server-cluster-ip-service spec: type: ClusterIP selector: component: server ports: - port: 5000 targetPort: 5000