Managed Kubernetes

• This is Kubernetes as a service offerings from cloud providers
• The managed k8s removes the burden of managing control plane and administrative activities are either automated or simplfied.
• Managed k8s is integrated with
  • cloud networks
  • cloud storages
• Managed k8s gives us the SLA ( Service Level Agreement) in terms of uptime and availability
• Popular Managed k8s
  • AKS (Azure Kubernetes Services)
  • EKS (Elastic Kubernetes Service)
  • GKE (Google K8s engine)
• Ensure kubectl is installed on your system
• How kubectl works?
  • kubectl needs a config file where the authentication information and server information are stored (~/.kube/config)

Azure AKS Creation

• Easiest way to create AKS is using Azure CLI
• Docs
• Steps

# variables
export MY_RESOURCE_GROUP_NAME="myAKSResourceGroup"
export REGION="eastus"
export MY_AKS_CLUSTER_NAME="myAKSCluster"
export MY_DNS_LABEL="mydnslabel"

# creates a resource group
az group create \
    --name $MY_RESOURCE_GROUP_NAME \
    --location $REGION

# creates an aks cluster 
# note: this command i will be running on cloudshell as it generates ssh keys
az aks create \
    --resource-group $MY_RESOURCE_GROUP_NAME \
    --name $MY_AKS_CLUSTER_NAME \
    --node-count 2 \
    --generate-ssh-keys \
    --node-vm-size Standard_B2ms

# kubectl can be install with az aks install-cli

# to get the kubeconfig file
az aks get-credentials \
    --resource-group $MY_RESOURCE_GROUP_NAME \
    --name $MY_AKS_CLUSTER_NAME

• check the connectivity

kubectl get nodes

• To remove k8s cluster

# delete the resource group
az group delete \
    --name $MY_RESOURCE_GROUP_NAME \
    --yes \
    --no-wait

Lets create a service of type load balancer in aks

• service of type lb and httpd rs
• Refer Here for the changes done to add service of type lb
• We have used a replicaset Refer Here
  

Health Probes in K8s

• official docs
• Three types of Health Probes/checks
  • Starup Probe:
    • Why: This verifies if the contianer is started or not.
    • What it does: if this check fails the container will be restarted
  • Liveness
    • Why: If the application is live or not (liveness)
    • What is does: restarts on failure
  • Readiness:
    • Why: If the application is ready to serve traffic
    • What it does: This pod will not recieve traffic from service on failure
• How to do the checks
  • exec: This check will run a command inside container and relies on exit code (0 => success all other numbers is failure)
  • tcp: This check will verify if the port is open for connection or not, Generally used for databases, cache servers
  • http: This check will send an http request to an endpoint and if it replies with 2xx or 3xx is considered as success
  • grpc: Here we can verify the grpc endpoint is responding

Lets change the spring petclinic replica set to have all the three probes

• image: shaikkhajaibrahim/febspc:1
• port: 8080
• Refer Here for replicaset spec