Learn containers!
IMPORTANT FOR ALL THE THINGS I use a macOS that has darwin ARM64... for most people they can use the minikube ip with node port to access things on their browser... not us lucky mac darwin users! instead heres what we have to do.
How to tunnel a service!
- get the service name by using
kubectl get svc - run
minikube service prometheus-server-ext - Wait a moment, it will open a tunnel and pop open the browser window for you.
Important General Knowledge
---------Main Differences Between Docker and Kubernetes-------- Docker = container platform K8s = container orechstration --> allows you to build your docker image and offers auto healing. auto scaling, clustering, load balancing, etc.
---------Main Differences Between Docker Swarm and Kubernetes--------
- dark swarm is only for simple aplications, simple and fast set up, support is more limited
- kubernetes is suitable for large orgs, offers more scalability, network capabilities policies, huge 3rd party ecosystem support
---------Docker container vs Kubernetes Pod---------
- docker container has the app itself packaged up and whatever is needed to run the appk
- kubernetes pod is defined using YAML (runtime spec) that lets you RUN 1 or more containers
---------What is a namespace in k8s?--------------
- its kind of like a project in AWS or GCP --> k8s = reviewly && workspaces = project1, project2
- lets you seperate workspaces, allow for multiple teams/projects without stomping on eachothers toes
- but its all the same cluster technically
--------day to day activities of DevOps eng on k8s--------------
- manage k8s clusters for org
- ensure apps are deployed and there are no issues with them
- monitoring is good
- no bugs
- maintenance activities
- upgrade versions, installing default mandatory packages, security updates
- subject matter exports
---------Main Components of k8 Architecture--------
- API Server --> handle apis that talks to end users.
- Scheduler --> schedules resources on the cluster
- Controller Manager (CM) --> manages the other controllers
- Cloud Controller Manager )CCM) --> has logic to spin up load balancers and stuff like that through cloud, basically how the cloud interfaces with k8s
- ETCD --> object store, all resources of k8s stored here as key value pairs
- Kubelet --> responsible to manage the containers within the pod the kubelet is on --> manage containers scheduled to run on the pod --> ensures healthy and running --> ensures proper resources they need are available --> communicates iwth k8 API --> gets info about containers that SHOULD be running on the node --> using this info it starts/stops containers as neeed --> restarts containers if need be
- Kube-Proxy --> networking, takes care of updating ip tables --> maintain network rules on each node --> updates roles dynamially as services added/removed --> client sends a request to a service, req intercepted by kubeproxy on node where it was received, kube-proxy llooks up destination endpoint for the service, routes request accordingly --> ensures services can communicate with eachother
- Container Runtime --> needed to run the actual container --> ex/ containerd
----What are labels and selectors used for?------
- used by service to identify pods
- even if a pod is deleted and a new pod is spun up the ip will change by label wont.
--------Different types of services in k8s---------
- cluster ip mode --> only exposed within cluster --> the cluster ip is only available within the cluster
- node port mode --> expose within network --> anyone in your org who has access to your node address
- load balance mode --> expose to the pubic --> by default can only be done on cloud
------More detail on difference between NodePort and LoadBalancer type service----
-
node port --> kube-proxy updates IPTables with node ip address & port choesen in the service config to access the pods
-
LB --> C-C-M creates external LB IP using underlying cloud provider login in the CCM and then users can access services using the external IPvb