DEV Community

Cover image for Certified Kubernetes Administrator (CKA) Prep.
Cosmas Nyairo
Cosmas Nyairo

Posted on

Certified Kubernetes Administrator (CKA) Prep.

#kubernetes #cka #devops #opensource

Hey there,

Introduction

For my CKA preparation, I have noted down topics i studied for in preparation for the exam as well as commands I use on a day to day basis.

N/B:

  • Most of the commands here I run them on a cloud9 instance whitelisted to access to my test EKS cluster.
  • A master/worker node setup on ec2 would work as well. Will share a guide on how that can be achieved.

Table of Contents

Notes

Useful commands and aliassed for the terminal:

# in .bashrc
alias k="kubectl"
export drc="--dry-run=client -oyaml"
export drs="--dry-run=server -oyaml"
export kre="kubectl replace --force -f"
# in .vimrc
set nu
set autoindent
Enter fullscreen mode Exit fullscreen mode 
kubectl explain po.spec
Enter fullscreen mode Exit fullscreen mode
  • etcd - key store for information about the cluster i.e nodes, pods, roles, secrets
  • kube-api sever - management component in kubernetes. Only component to interact with the etcd data store
  • kube-controller-manager - manages contollers in kubernetes and brings the system to the desired state
  • kube-scheduler - decides which pod goes to which node
  • kubelet - registers nodes, creates pods on a nodes, monitors the pods and nodes
  • kube-proxy - creates forwarding rules for nodes in the cluster 
If we install the cluster using kubeadmin, we can view :
  - Exec into the etcd pod and run the `etcdctl` command
  - Exec into the kube-apiserver pod and access the `/etc/kubernetes/manifests/kube-apiserver.yaml` file
  - Exec into the kube-controller manager pod and access the `/etc/kubernetes/manifests/kube-controller-manager.yaml` file
  - Exec into the kube-scheduler pod and access the `/etc/kubernetes/manifests/kube-scheduler.yaml` file
  - Kubeadmin doesn't deploy the kubelet, we have to do it manually
Enter fullscreen mode Exit fullscreen mode 
If we install the clustrer without using kubeadmin, we can view:
  - explore the etcd service
  - the apiserver service located at `/etc/systemd/system/kube-apiserver.service` or `ps -aux | grep kube-apiserver` on the master node.
  - the kubecontroller manager service located at `/etc/systemd/system/kube-controller-manager.service` or `ps -aux | grep kube-controller-manager` on the master node.
  - explore the kube-scheduler service or `ps -aux | grep kube-scheduler`
  - explore the kubelet service or `ps -aux | grep kubelet`
Enter fullscreen mode Exit fullscreen mode

  • Kubelet config is at: /var/lib/kubelet/config.yaml for each node

  • Test commands

kubectl run webapp-green --image=test --dry-run=server -oyaml
Enter fullscreen mode Exit fullscreen mode 
kubectl auth can-i
Enter fullscreen mode Exit fullscreen mode
  • To pass commands to pods on creation: 
kubectl run webapp-green --image=test --command -- --color=green
Enter fullscreen mode Exit fullscreen mode
  • To pass args to pods on creation: 
 kubectl run webapp-green --image=test -- --color=green
Enter fullscreen mode Exit fullscreen mode

Update env variables:

kubectl -n test set env deploy/test --containers=container-name DB_Host=mysql DB_User=user1 DB_Password=pass123
Enter fullscreen mode Exit fullscreen mode

Check certificate:

openssl x509 -in /opt/cert.crt -text
Enter fullscreen mode Exit fullscreen mode

Check service logs:

journalctl -u service
Enter fullscreen mode Exit fullscreen mode 
service servicename status
Enter fullscreen mode Exit fullscreen mode

Kubelet file located at: var/lib/kubelet/config.yaml

Scheduling

Manual scheduling

  • Add the nodeName field when creating a pod only during creation time
  • We can alternatively create a pod binding file as shown below: 
apiVersion: v1
kind: Binding
metadata:
  name: myapp
target:
  apiVersion: v1
  kind: Node 
  name: my-node
Enter fullscreen mode Exit fullscreen mode
  • Then, we send a post request to the pod's binding api with data section set to the binding object in a json format as follows: 
curl --header "Content-Type:application/json" --request POST --data <<BINDING-OBJECT-YAML>> http://$SERVER/api/v1/namespaces/<namespace>/pods/$PODNAME/binding/
Enter fullscreen mode Exit fullscreen mode

Labels and Selectors

  • Annotations record other details in pod for informatory purposes 
  kubectl get po --selector type=test
Enter fullscreen mode Exit fullscreen mode

Taints and Tolerations

  • Restrict what pods can be scheduled on a node
  • Taints set on nodes, tolerations set on pods
  • Taint effect = what happens to pods that can't tolerate the taint
    • NoSchedule
    • PreferNoSchedule
    • NoExecute

Taint and toleration example yaml shown below:

apiVersion: v1
kind: Pod
metadata:
  name: <<pod-name>>
spec:
  containers:
    - name: nginx-container
      image: nginx
  tolerations:
    - key: "type"
      value: "test"
      operator: "Equal"
      effect: "NoSchedule"
Enter fullscreen mode Exit fullscreen mode 
  kubectl taint nodes my-node type=test:NoSchedule
Enter fullscreen mode Exit fullscreen mode 
  kubectl taint node my-node node-role=kubernetes.io/control-plane:NoSchedule-
Enter fullscreen mode Exit fullscreen mode

Node Selector

  • Add node selector section in the pod with the label of the node we want to use as shown below: 
apiVersion: v1
kind: Pod
metadata:
  labels:
    name: myapp
spec:
  containers:
  - name: myapp
    image: busybox
    resources:
      limits:
        memory: "128Mi"
        cpu: "500m"
    ports:
      - containerPort: 80
  nodeSelector:
    type: large
Enter fullscreen mode Exit fullscreen mode 
  kubectl label nodes my-node type=test
Enter fullscreen mode Exit fullscreen mode

Node affinity

Node affinity example below:

apiVersion: v1
kind: Pod
metadata:
  name: <<pod-name>>
  labels:
    app: <<label-value>>
    type: <<label-value>>
spec:
  containers:
    - name: nginx-container
      image: nginx
  affinity:
    nodeAffinity:
      requiredDuringSchedulingIgnoredDuringExecution:
        nodeSelectorTerms:
          - matchExpressions:
              - key: type
                operator: Exists
          # - matchExpressions:
          #     - key: type
          #       operator: In
          #       values:
          #         - test
Enter fullscreen mode Exit fullscreen mode

Node affinity types:

  • requiredDuringSchedulingIgnoredDuringExecution
  • PreferredDuringSchedulingIgnoredDuringExecution
  • requiredDuringSchedulingrequiredDuringExecution
During Scheduling During Execution
1 Required Ignored
2 Preferred Ignored
3 Required Required

We can combine node affinity with taints and tolerations

Resource Limits

We can create limit ranges which need to be followed by pods in a namespace as shown below:

apiVersion: v1
kind: LimitRange
metadata:
  name: cpu-limit-range
spec:
  limits:
    - default:
        cpu: 1
      defaultRequest:
        cpu: 0.5
      type: Container
---
apiVersion: v1
kind: LimitRange
metadata:
  name: mem-limit-range
spec:
  limits:
    - default:
        memory: 512Mi
      defaultRequest:
        memory: 256Mi
      type: Container
---
Enter fullscreen mode Exit fullscreen mode

Daemon Sets

Ensures a copy of a pod is always on each node in the cluster. Use case: Logging, monitoring agents as shown below:

apiVersion: apps/v1
kind: DaemonSet
metadata:
  name: mydaemonset
spec:
  selector:
    matchLabels:
      app: mydaemonset-app
  template:
    metadata:
      labels:
        app: mydaemonset-app
    spec:
      containers:
      - name: myapp
        image: logviewer
Enter fullscreen mode Exit fullscreen mode

Static Pods

Pods created independently by the kubelet without help from the kubeapiserver.
We can do this in the following ways:

  • find the kubelet service config path:

    ps -aux | grep -i kubelet | grep -i config

  • check the node's /var/lib/kubelet/config.yaml path

  • Configure the pod definition path in the kubelet service at with the following entry: --pod-manifest-path=etc/kubernetes/manifests

  • Provide a --config=mydefinition.yaml entry in the kubelet service and have the mydefinition.yaml file contain the following:

      staticPodPath: etc/kubernetes/manifests

  • For both approaches we then place the pod definition files at the specified directory in our case it's /etc/kubernetes/manifests and view pods created via the docker ps command

We can use static pods to deploy control plane components as static pods.

Multiple Schedulers

We can create our custom scheduler in kubernetes then pass the file to the custom scheduler in this format:
--config=/etc/kubenetes/my-scheduler/my-scheduler-conf.yaml.

The my-scheduler-conf.yaml file contains the following:

apiVersion: kubescheduler.config.k8s.io/v1beta2
kind: KubeSchedulerConfiguration
profiles:
  - schedulerName: my-scheduler
leaderElection:
  leaderElect: true
  resourceNamespace: kube-system
  resourceName: lock-object-my-scheduler
Enter fullscreen mode Exit fullscreen mode

The leaderelection section is used when we have multiple master nodes with the custom scheduler running on them in a HA(High Availability) setup and we need to chose the active copy of the scheduler to be running at a time.

Read more at: Multiple Schedulers

For pods to use our custom scheduler, we provide the scheduler name field in the pod definition.

apiVersion: v1
kind: Pod
metadata:
  name: myapp
  labels:
    name: myapp
spec:
  containers:
  - name: myapp
    image: busybox
    resources:
      limits:
        memory: "128Mi"
        cpu: "500m"
    ports:
      - containerPort: 2000
  schedulerName: my-custom-scheduler
Enter fullscreen mode Exit fullscreen mode

Scheduler Profiles

We can disable existing scheduler plugins as well as enable our own custom plugins as below

apiVersion: kubescheduler.config.k8s.io/v1beta2
kind: KubeSchedulerConfiguration
profiles:
  - schedulerName: my-scheduler
    plugins:
      filter:
        disabled:
          - name: TaintToleration
        enabled:
          - name: MyCustomPlugin
  - schedulerName: my-scheduler-two
    plugins:
      score:
        disabled:
          - name: 'NodeAffinity'
          - name: 'ImageLocality'
Enter fullscreen mode Exit fullscreen mode

Monitoring and Logging

We can use the metric server from: metrics-server to collect metrics about our cluster performance. We can also used advanced tools for out monitoring i.e prometheus using this guide

We can then view performance using

kubectl top node
Enter fullscreen mode Exit fullscreen mode 
kubectl top pods
Enter fullscreen mode Exit fullscreen mode

To get streamed logs of a currently running pod we use

kubectl logs pod-name -f
Enter fullscreen mode Exit fullscreen mode

To get streamed logs of a currently running pod with multiple containers we use

kubectl logs pod-name -c container-name -f
Enter fullscreen mode Exit fullscreen mode

Os Upgrades

  • We can drain a node to remove the pods currently running on it and then the node is cordoned (marked as unschedulable) 
kubectl drain node1
Enter fullscreen mode Exit fullscreen mode
  • We can also manually cordon a node 
kubectl cordon node1
Enter fullscreen mode Exit fullscreen mode
  • After our upgrade/fixes to the node, we then uncordon the node once it's back up 
kubectl uncordon node1
Enter fullscreen mode Exit fullscreen mode
  • We can have the different components with different versions but not a higher version than the kube api-server but kubectl can be one version higher or lower or the same version
Component Maximum version
Kube Api Server x
Controller Manager x-1
Kube Scheduler x-1
Kubectl x-1 or x or x+1
Kubelet x-2
Kube-Proxy x-2
  • To upgrade a cluster managed by kubeadm,
    • we first start with the master node then the worker nodes.
    • We first get the upgrade plan for our version,

Step 1 (master node):

kubeadm upgrade plan
Enter fullscreen mode Exit fullscreen mode 
apt-get upgrade kubeadm=<version> -y
Enter fullscreen mode Exit fullscreen mode 
kubeadm version
Enter fullscreen mode Exit fullscreen mode 
kubeadm upgrade apply <version>
Enter fullscreen mode Exit fullscreen mode

If we have kubelets on the master node:

apt-get upgrade kubelet=<version> -y
Enter fullscreen mode Exit fullscreen mode 
systemctl restart kubelet
Enter fullscreen mode Exit fullscreen mode

Step 2 (worker node-> one by one):

kubectl drain node01 # on master node
Enter fullscreen mode Exit fullscreen mode 
apt-get upgrade kubeadm=<version> -y
Enter fullscreen mode Exit fullscreen mode 
apt-get upgrade kubelet=<version> -y
Enter fullscreen mode Exit fullscreen mode 
systemctl restart kubelet
Enter fullscreen mode Exit fullscreen mode 
kubectl uncordon node01  # on master node
Enter fullscreen mode Exit fullscreen mode

Backup candidates : Resource configurations, etcd

  • For resource configurations: 
  kubectl get all --all-namespaces -o yaml > deploy-and-service.yaml
Enter fullscreen mode Exit fullscreen mode
  • For Etcd, We can either backup the directory where etcd stores to or take a snapshot and restore it 
  export ETCDCTL_API=3
Enter fullscreen mode Exit fullscreen mode 
  etcdctl snapshot save --endpoints=<enpoints> \
                        --cacert=<path-to-cacert> \
                        --cert=<path-to-cert> \
                        --key=<path-to-key>
                        /path/to/backup/file/mysnapshot.db
Enter fullscreen mode Exit fullscreen mode 
  etcdctl snapshot status /path/to/backup/file/mysnapshot.db
Enter fullscreen mode Exit fullscreen mode 
  service kube-api-server stop
Enter fullscreen mode Exit fullscreen mode 
  etcdctl snapshot restore /path/to/backup/file/mysnapshot.db --data-dir /path/to/restore/to/
  # We then update etcd configuration to use the new directory
Enter fullscreen mode Exit fullscreen mode 
  service etcd restart
  service kube-api-server start
Enter fullscreen mode Exit fullscreen mode

We can use scp to copy objects to a pod:

scp test.txt my-server:/home
Enter fullscreen mode Exit fullscreen mode 
scp my-server:/home test.txt
Enter fullscreen mode Exit fullscreen mode

To join cluster:

kubeadm token create --print-join-command
Enter fullscreen mode Exit fullscreen mode

More info: Disaster Recovery for Kubernetes Clusters

Security

Authentication

Authentication to the kube-api server is through:

  • Password file:

    • Specify in kubeapiserver service or kubeapi server yaml file --basic-auth-file=usercreds-basic.csv
    • The csv file should be in the fomat of: 
    password,username,userid,groupid
    • If using api to authenticate pass via curl -v -k https://masternod-ip:port/api/vi/pods -u "user1:password1"

  • Token file:

    • Specify in kubeapiserver service or kubeapi server yaml file --token-auth-file=usercreds-token.csv
    • The csv file should be in the fomat of: 
    token,username,userid,groupid
    • If using api to authenticate pass as a bearer token curl -v -k https://masternod-ip:port/api/vi/pods --header "Authorization: Bearer <TOKEN>"

  • Certificates

  • External Identity Service

TLS Certificates

Cerificates with public key usually end with .crt or .pem while those with private keys usually end with .key or -key.pem

List of certificates in the cluster

  • Certificate Authority certs (we can use more than one ca)
  • Server certificates:
    • kube-api-server
    • etcd server
    • kubelet server
  • Client certificates:
    • users (admins)
    • kube-scheduler
    • kube-controller-manager
    • kube-proxy
    • kube-api-server to kubelet
    • kube-api-server to etcd

All certificate operations done by the kube controller manager. We specify the following:

    - --cluster-signing-cert-file='pathtofile'
    - --cluster-signing-key-file='pathtofile'
Enter fullscreen mode Exit fullscreen mode

Certificate signing requests using the following file:

cat file.csr | base64 -w 0
Enter fullscreen mode Exit fullscreen mode 
apiVersion: certificates.k8s.io/v1
kind: CertificateSigningRequest
metadata:
  name: johndoe
spec:
  request: BASE64 ENCODED CSR
  signerName: kubernetes.io/kube-apiserver-client
  groups: 
    - system:authenticated
  expirationSeconds: 86400  # one day
  usages:
  - server auth
  - key encryptment
Enter fullscreen mode Exit fullscreen mode 
kubectl get csr
Enter fullscreen mode Exit fullscreen mode 
kubectl certificate approve csr-name
Enter fullscreen mode Exit fullscreen mode 
kubectl certificate deny csr-name
Enter fullscreen mode Exit fullscreen mode 
kubectl get csr csr-name -o yaml
Enter fullscreen mode Exit fullscreen mode

Kubeconfig

We can send request using kubectl command without kubeconfig as follows:

kubectl get pods --server server --client-key client.key --client-certificate client.crt --certificate-authority ca.crt
Enter fullscreen mode Exit fullscreen mode

We can use a kubeconfig file to remove the need to add the details and send a request like:

kubectl get pods
Enter fullscreen mode Exit fullscreen mode

Kubeconfig file has these sections:

  • Clusters (dev,prod)
  • Users (developer, admin, qa)
  • Contexts (user-to-cluster i.e developer@dev, admin@prod)

Sample kubeconfig file:

apiVersion: v1
kind: Config
current-context: developer@localcluster-context 
clusters:
- name: remotecluster
  cluster:
    certificate-authority: path/remote-ca.crt
    server: https://ip:port
- name: localcluster
  cluster:
    certificate-authority-data: BASE64 ENCODED CSR
    server: https://ip:port
contexts:
- name: kube-admin@localcluster-context 
  context:
    cluster: remotecluster
    user: kube-admin
    namespace: kube-system
- name: developer@localcluster-context 
  context:
    cluster: localcluster
    user: developer
users:
- name: kube-admin
  user:
    client-certificate: path/admin-client.crt
    client-key: path/client.key
- name: developer
  user:
    client-certificate: path/dev-client.crt
    client-key: path/client.key
Enter fullscreen mode Exit fullscreen mode 
kubectl config view 
kubectl config view --kubeconfig path-to-config
Enter fullscreen mode Exit fullscreen mode 
kubectl cluster-info --kubeconfig path-to-config
Enter fullscreen mode Exit fullscreen mode 
kubectl config use-context developer@localcluster-context
Enter fullscreen mode Exit fullscreen mode 
kubectl config set-credentials user --client-certificate=/path/user.crt
Enter fullscreen mode Exit fullscreen mode

Authorization

We set authorization mode in the kubeapiserver entry --authorization-mode=Node,RBAC and they are used following the order they were specified

Role file:

apiVersion: rbac.authorization.k8s.io/v1
kind: Role
metadata:
  name: developer
rules:
  - apiGroups: [""]
    resources: ["pods"]
    verbs: ["get", "watch", "list"]
    resourceNames:
      - "devhour"
  - apiGroups: ["apps"]
    resources: ["deployments"]
    verbs: ["get", "list", "watch"]
Enter fullscreen mode Exit fullscreen mode

Role binding file:

apiVersion: rbac.authorization.k8s.io/v1
kind: RoleBinding
metadata:
  name: developer-binding
subjects:
- kind: Group
  name: developers
  apiGroup: rbac.authorization.k8s.io
roleRef:
  kind: Role #this must be Role or ClusterRole
  name: developer # this must match the name of the Role or ClusterRole you wish to bind to
  apiGroup: rbac.authorization.k8s.io
Enter fullscreen mode Exit fullscreen mode 
kubectl get roles 
kubectl get rolebindings
Enter fullscreen mode Exit fullscreen mode 
kubectl auth can-i get pods
Enter fullscreen mode Exit fullscreen mode 
kubectl auth can-i get pods --as developer --namespace dev
Enter fullscreen mode Exit fullscreen mode 
kubectl create role rolename --verb=create,list,delete --resource=pod
Enter fullscreen mode Exit fullscreen mode 
kubectl create rolebinding rolebindingname --role=rolename --user=user-to-bind
Enter fullscreen mode Exit fullscreen mode

To view namespaced resources: --namespaced=true
We can use cluster roles on namespaced resources to give access to the objects in the whole cluster

Cluster Role file:

apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRole
metadata:
  name: cluster-admin
rules:
  - apiGroups: [""]
    resources: ["nodes"]
    verbs: ["get", "watch", "list"]
Enter fullscreen mode Exit fullscreen mode

Cluster Role binding file:

apiVersion: rbac.authorization.k8s.io/v1
kind: ClusterRoleBinding
metadata:
  name: admin-binding
subjects:
- kind: Group
  name: admin
  apiGroup: rbac.authorization.k8s.io
roleRef:
  kind: ClusterRole #this must be Role or ClusterRole
  name: cluster-admin # this must match the name of the Role or ClusterRole you wish to bind to
  apiGroup: rbac.authorization.k8s.io
Enter fullscreen mode Exit fullscreen mode 
kubectl create clusterrole clusterrolerolename --verb=* --resource=pod
Enter fullscreen mode Exit fullscreen mode 
kubectl create clusterrolebinding clusterrolebindingname --clusterrole=clusterrolerolename --user=user-to-bind
Enter fullscreen mode Exit fullscreen mode

For service accounts:

kubectl create sa service-account-name`
Enter fullscreen mode Exit fullscreen mode 
kubectl create token service-account-name
Enter fullscreen mode Exit fullscreen mode

For registry secrets:

 kubectl create secret docker-registry NAME --docker-username=user --docker-password=password --docker-email=email --docker-server=string
Enter fullscreen mode Exit fullscreen mode

Network Policy

By default kubernetes allows traffic from all pods to all destinations.
We can define egress and ingress rules to block traffic to and from pods.

If we add ingress rule, we don't need to specify the egress rule for the response as it is allowed automatically.
We do need an egress rule if we need to send external api calls

apiVersion: networking.k8s.io/v1
kind: NetworkPolicy
metadata:
  name: network-policy-name
spec:
  podSelector:
    matchLabels:
      test: testing # pods matching this label will have the network policy applied on them
  policyTypes:
    - Ingress
    - Egress
  ingress:
    - from:
        # The pod selector and namespace selector rules both need to be met.
        # We can add a - before the namespace selector to make it that either of them need to be met.
        - podSelector:
            matchLabels:
              name: pod-name # allow pods with this label to send ingress traffic
          namespaceSelector:
            matchLabels:
              name: testing # specify pods from which namespace are allowed to to send ingress traffic
        - ipBlock:
            cidr: ip-cidr  # range of ip addresses allowed to send ingress traffic
        ports:
          - protocol: "TCP"
            port: 3306 # port to recieve ingress traffic
  egress:
      - to:
          - podSelector:
              matchLabels:
                name: pod-name  # send egress traffic to pods with this label 
            namespaceSelector:
              matchLabels:
                name: testing # send egress traffic to pods in the namespace with this label
          - ipBlock:
            cidr: ip-cidr  # send egress traffic to this range of ip addresses
          ports:
            - protocol: "TCP"
              port: 80 # port to send egress traffic to
Enter fullscreen mode Exit fullscreen mode

Storage

To create a volume in a pod:

apiVersion: v1
kind: Pod
metadata:
  name: <<pod-name>>
  labels:
    app: <<label-value>>
    type: <<label-value>>
spec:
  containers:
    - name: nginx-container
      image: nginx
      volumeMounts:
        - mountPath: /opt
          name: test-volume
        - mountPath: /home/
          name: ebs-volume
  volumes:
    - name: test-volume
      hostPath:
        path: /home/data
        type: Directory
    - name: ebs-volume
      awsElasticBlockStore:
        volumeID: <<vol-id>>
        fsType: "ntfs"
Enter fullscreen mode Exit fullscreen mode

Persistent Volumes: A cluster wide pool of storage volumes which we can assign to applications on the cluster.

apiVersion: v1
kind: PersistentVolume
metadata:
  name: persistent-vol
spec:
  capacity:
    storage: 1Gi
  accessModes:
    - ReadWriteOnce
  persistentVolumeReclaimPolicy: Retain
  awsElasticBlockStore:
    volumeID: <<vol-id>>
    fsType: "ntfs"
Enter fullscreen mode Exit fullscreen mode

Persistent Volumes Claim: Select storage from the persistent volumes created.

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: persitent-vol-claim
spec:
  resources:
    requests:
      storage: 500Mi
  accessModes:
    - ReadWriteOnce
Enter fullscreen mode Exit fullscreen mode

Using Persistent Volumes Claim in a pod:

apiVersion: v1
kind: Pod
metadata:
  name: <<pod-name>>
  labels:
    app: <<label-value>>
    type: <<label-value>>
spec:
  containers:
    - name: nginx-container
      image: nginx
      volumeMounts:
        - mountPath: /opt
          name: pvc-volume
  volumes:
    - name: pvc-volume
      persistentVolumeClaim:
        claimName: persitent-vol-claim
Enter fullscreen mode Exit fullscreen mode

Storage Classes: We don't need to creat a persitent volume. This is done automatically

apiVersion: storage.k8s.io/v1
kind: StorageClass
metadata:
  name: ebs-storage
provisioner: kubernetes.io/aws-ebs
parameters:
  type: gp2
  fsType: ext4
  encrypted: true
Enter fullscreen mode Exit fullscreen mode

Persitent volume storage class:

apiVersion: v1
kind: PersistentVolumeClaim
metadata:
  name: persitent-vol-claim
spec:
  resources:
    requests:
      storage: 500Mi
  storageClassName: ebs-storage
  accessModes:
    - ReadWriteOnce
Enter fullscreen mode Exit fullscreen mode

Networking

Network configuration - Cluster Nodes

Ports required for kubernetes: Ports

Find network interfaces:

ip a
Enter fullscreen mode Exit fullscreen mode

Find bridges in network:

ip a show type bridge
Enter fullscreen mode Exit fullscreen mode

List routes in the host

ip route
Enter fullscreen mode Exit fullscreen mode

Get active internet connections

netstat -nplt
Enter fullscreen mode Exit fullscreen mode

Get number of client connections on a program running on a port

netstat -anp | grep 'etcd\|2379' | wc -l
Enter fullscreen mode Exit fullscreen mode

Pod Networking

When a container is created, the kubelet looks at the cni configuration passed:

  • network-plugin=cni
  • cni-conf-dir=/etc/cni/net.d to find our scipts name
  • cni-bin-dir=/opt/cni/bin to find the script

The kubelet then runs the script with the add command with the name and namespce id of the container net-script.sh add <container> <namespace>

Service Networking

  • Cluster Ip services are not bound to a specific node but are available to all pods cluster wide
  • Node Port services exposes application on a port on all nodes in the cluster.

Kube proxy creates forward rules in all nodes such that when we try to reach the ip of a service, the request is forwarded to the ip and port of the pod.

For kubeproxy, we set the proxymode : --proxy-mode [userspace | iptables | ipvs] If not set, if defaults to iptables.

For services, we set the range in the kube-api-server --service-cluster-ip-range <range> which defaults to 10.0.0.0/24

View rules created by kube proxy

iptables -L -t nat | grep -i servicename
Enter fullscreen mode Exit fullscreen mode

DNS

  • Domain resolution for services: web-service.namespace.svc.cluster-root-domain i.e db-service.test-ns.svc.cluster.local
  • For pods , we replace the . with -. This is turned off by default.
  • Domain resolution for pods: pod-ip-replaced-with-hyphens.namespace.pod.cluster-root-domain i.e 10-244-1-2.test-ns.pod.cluster.local
  • Coredns watches the cluster for new pods or services and adds a record for them in it's database when they're created and it also creates a service named kube-dns.
  • Core file for coredns at : etc/coredns/Corefile and it's passed as a config map object
  • For kubelet, we have the ip of dns server and domain in its config

Example Corefile

{
    errors
    health {
       lameduck 5s
    }
    ready
    kubernetes cluster.local in-addr.arpa ip6.arpa {
       pods insecure
       fallthrough in-addr.arpa ip6.arpa
       ttl 30
    }
    prometheus :9153
    forward . /etc/resolv.conf {
       max_concurrent 1000
    }
    cache 30
    loop
    reload
    loadbalance
}
Enter fullscreen mode Exit fullscreen mode

To get the service full domain name:

host service-name
Enter fullscreen mode Exit fullscreen mode

Json Path

Kubernetes docs for json path: Kubernetes Jsonpath
Use the following evaluator for json JsonPath Evaluator

#get field names
k explain deploy.spec --recursive
Enter fullscreen mode Exit fullscreen mode 
# Get all car names in an json data is equal to "abc"
# $[] -> return a list
# ?() -> if 
# @ -> Each item in list

$[?(@.name == "abc")]
Enter fullscreen mode Exit fullscreen mode 
#Get name and image of first container of the first pod 
kubectl get pods -o=jsonpath='{range .items[0].spec.containers[0]}{.name}{"\n"}{.image}' --sort-by=.name
Enter fullscreen mode Exit fullscreen mode 
#Fetch node names 
kubectl get nodes -ojsonpath='{range .items[*].metadata}{.name}
Enter fullscreen mode Exit fullscreen mode 
# Sort persisten volumes by capacity
kubectl get pv --sort-by=.spec.capacity.storage
Enter fullscreen mode Exit fullscreen mode 
kubectl get pv --sort-by=.spec.capacity.storage -o=custom-columns=NAME:.metadata.name,CAPACITY:.spec.capacity.storage
Enter fullscreen mode Exit fullscreen mode 
# get names of user test in config
kubectl config view --kubeconfig=custom-config -o jsonpath="{.contexts[?(@.context.user=='test')].name}"
Enter fullscreen mode Exit fullscreen mode 
#Get deployments as per this format:
# DEPLOYMENT   CONTAINER_IMAGE   READY_REPLICAS   NAMESPACE

kubectl get deploy -n admin2406 --sort-by=.metadata.name -o=custom-columns=DEPLOYMENT:.metadata.name,CONTAINER_IMAGE:.spec.template.spec.containers.*.image,READY_REPLICAS:.status.readyReplicas,NAMESPACE:.metadata.namespace
Enter fullscreen mode Exit fullscreen mode 
kubectl get svc --sort-by=.metadata.name  -o=custom-columns=NAME:.metadata.name,PORTNAME:.spec.ports.name
Enter fullscreen mode Exit fullscreen mode 
k get nodes -ojsonpath='{range .items[*]}InternalIP of {.metadata.name} {.status.addresses[?(@.type=="InternalIP")].address} {end}'
Enter fullscreen mode Exit fullscreen mode

Top comments (0)

Read next

samchon profile image

I made Swagger/OpenAPI and LLM function calling schema definitions

Jeongho Nam -

abhay_yt_52a8e72b213be229 profile image

Docker on Kubernetes (K8s): A Powerful Combo for Scalable Containerized Applications

Abhay Singh Kathayat -

ssheffk profile image

🚀 Learning by Doing: Building an Incident Alert System 🛠️

Serdar Shefketov -

abhay_yt_52a8e72b213be229 profile image

Top 100 Kubernetes Topics to Master for Developers and DevOps Engineers

Abhay Singh Kathayat -