Understanding Deployments

The Deployment is the standard for running applications in Kubernetes
It offers features that add to the scalability and reliability of the application
Offers scalability scaling the number of application instances
Updates and Update Strategy: zero-downtime application updates
It protects Pods and will automatically restart them if anything goes wrong
Use kubectl create deploy to create Deployments
Replication Controller & Replica Set is not able to do update & Rollback apps in the cluster.
We have a restart policy but that's for containers not for pods. So if we delete a pod it's gone not gonna start again.

For Exmaple

kubectl create deployment --image=nginx --replicas=3

The default label will be the name of deployment itself.

Overview

Deployment object act as a supervisor for pods, giving you free-grained control over how and when a new pod is rolled out, updated or rolled back to a previous state.
When using deployment object, we first define the state of the app, then Kubernetes cluster schedules mentioned app instance on the specific nodes
Kubernetes then monitors, if the node hosting an instance goes down or pod is deleted, the deployment controller replaces it.
This provides a self hosting mechanism to address machine failure or maintenance.

Use cases of Deployment object

Create deployment to roll out a replica set - the replica set creates pods in the background. Check the status of the rollout to see if it succeeds or not
Declare the new state of the pods - By updating the podTemplateSpec of the deployment. A new replica set is created and the deployment manages moving the pods from the old replica set to the new one at a controlled rate. Each new replica set update the revision of the deployment.
Rollback to an earlier deployment revision - If the current state of the deployment is not stable. Each rollback updates the revision of the deployment
Scale up the deployment to facilitates more load.
Pause the deployment to apply multiple fixes to its podTemplateSpec and then resume it to start a new rollout.
Clean up older ReplicaSets that you don't need anymore
If there are problems in the deployment, Kubernetes will automatically roll back to the previous version, however you can also explicitly rollback to a specific revision as in our case to revision 1 (the original pod version).

Deployment Failure cases

Your deployment may get stuck trying to deploy its newest ReplicaSet without ever completing. This can occur due to some of the following factors -

Insufficient quota
Readiness probe failure
Image pull error
Insufficient permission
Limit ranges
Application runtime missing

Implementation

You can rollback to a specific version by specifying it with --to-revision

For example -

kubectl rollout undo deploy/my-deployment --to-revision=2

The name of the replicaset and pod is always formatted as [DEPLOYMENT_NAME]-[REPLICASET (Random string)]-[POD-NAME (Random string)]

When you display the inspect of deployment the following fields are displayed -

Field	Description
`NAME`	List the names of the deployment in the namespace
`READY`	Display how many replicas of application are available to your users if follows the pattern `Ready/Desired`
`UP TO DATE`	Display the number of replicas that have been updated to achieve the desired state
`AVAILABLE`	Display how many replicas of the application are available to your users
`AGE`	Display the amount of time that the application has been running

Example of Deployment8

k8s_example_9.yml

kind: Deployment
apiVersion: apps/v1
metadata:
  name: example-9
spec:
  replicas: 2
  selector:
    matchLabels:
      name: xander
  template:
    metadata:
      name: container0
      labels:
        name: xander
    spec:
      containers:
        - name: c00
          image: ubuntu
          command: [
              # Use a list for the command
              "/bin/bash",
              "-c",
              "while true; do echo Hello-from-ubuntu; sleep 5 ; done",
            ]

Create a deployment

kubectl apply -f k8s_example_9.yml

Output

controlplane $ kubectl apply -f k8s_example_9.yml
deployment.apps/example-9 created

Details of deployment

kubectl describe deployment example-9

Output

controlplane $ kubectl describe deployment example-9
Name:                   example-9
Namespace:              default
CreationTimestamp:      Wed, 16 Oct 2024 17:07:03 +0000
Labels:                 <none>
Annotations:            deployment.kubernetes.io/revision: 1
Selector:               name=xander
Replicas:               2 desired | 2 updated | 2 total | 2 available | 0 unavailable
StrategyType:           RollingUpdate
MinReadySeconds:        0
RollingUpdateStrategy:  25% max unavailable, 25% max surge
Pod Template:
  Labels:  name=xander
  Containers:
   c00:
    Image:      ubuntu
    Port:       <none>
    Host Port:  <none>
    Command:
      /bin/bash
      -c
      while true; do echo Hello-from-Ubuntu; sleep 5 ; done
.
.
NewReplicaSet:   example-9-6ddd44564c (2/2 replicas created)
Events:
  Type    Reason             Age   From                   Message
  ----    ------             ----  ----                   -------
  Normal  ScalingReplicaSet  15m   deployment-controller  Scaled up replica set example-9-6ddd44564c to 2

Details of replicaset created by deployment

kubectl get rs

Output

controlplane $ kubectl get rs
NAME                   DESIRED   CURRENT   READY   AGE
example-9-6ddd44564c   2         2         2       18m

Details of pods created by deployment in the replicaset

kubectl get pods

Output

controlplane $ kubectl get pods
NAME                         READY   STATUS    RESTARTS   AGE
example-9-6ddd44564c-c62ph   1/1     Running   0          19m
example-9-6ddd44564c-djbcd   1/1     Running   0          19m

Scaling using deployment

kubectl scale --replicas=1 deployment example-9
kubectl get deployment
kubectl get rs
kubectl get pods

Output

controlplane $ kubectl scale --replicas=1 deployment example-9
deployment.apps/example-9 scaled
controlplane $ kubectl get deployment
NAME        READY   UP-TO-DATE   AVAILABLE   AGE
example-9   1/1     1            1           22m
controlplane $ kubectl get rs
NAME                   DESIRED   CURRENT   READY   AGE
example-9-6ddd44564c   1         1         1       22m
controlplane $ kubectl get pods
NAME                         READY   STATUS        RESTARTS   AGE
example-9-6ddd44564c-c62ph   1/1     Running       0          22m
example-9-6ddd44564c-djbcd   1/1     Terminating   0          22m

Verify the OS of the pod

kubectl exec -it example-9-6ddd44564c-djbcd -- /bin/bash

Output

controlplane $ kubectl exec -it example-9-6ddd44564c-djbcd -- cat /etc/os-release
PRETTY_NAME="Ubuntu 24.04.1 LTS"
NAME="Ubuntu"
.
.
.

Trying Rollback

Let's create a new revision by changing the image type in the manifest. So, that we will have two versions of the same deployment and then try to rollout and rollback to previous version

k8s_example_9.yml

kind: Deployment
apiVersion: apps/v1
metadata:
  name: example-9
spec:
  replicas: 2
  selector:
    matchLabels:
      name: xander
  template:
    metadata:
      name: container0
      labels:
        name: xander
    spec:
      containers:
        - name: c00
          image: centos
          command: [
              # Use a list for the command
              "/bin/bash",
              "-c",
              "while true; do echo Hello-from-CentOS; sleep 5 ; done",
            ]

Create a deployment

kubectl apply -f k8s_example_9.yml

Output

controlplane $ kubectl apply -f k8s_example_9.yml
deployment.apps/example-9 created

Details of deployment

kubectl describe deployment example-9

Notice the changes compare the highlighted changes with the first revision

Output

controlplane $ kubectl describe deployment example-9
Name:                   example-9
Namespace:              default
CreationTimestamp:      Wed, 16 Oct 2024 17:07:03 +0000
Labels:                 <none>
Annotations:            deployment.kubernetes.io/revision: 2
Selector:               name=xander
Replicas:               2 desired | 2 updated | 2 total | 2 available | 0 unavailable
StrategyType:           RollingUpdate
MinReadySeconds:        0
RollingUpdateStrategy:  25% max unavailable, 25% max surge
Pod Template:
  Labels:  name=xander
  Containers:
   c00:
    Image:      centos
    Port:       <none>
    Host Port:  <none>
    Command:
      /bin/bash
      -c
      while true; do echo Hello-from-CentOS; sleep 5 ; done
    Environment:   <none>
    Mounts:        <none>
  Volumes:         <none>
  Node-Selectors:  <none>
  Tolerations:     <none>
Conditions:
  Type           Status  Reason
  ----           ------  ------
  Available      True    MinimumReplicasAvailable
  Progressing    True    NewReplicaSetAvailable
OldReplicaSets:  example-9-6ddd44564c (0/0 replicas created)
NewReplicaSet:   example-9-6755695d56 (2/2 replicas created)
Events:
  Type    Reason             Age                    From                   Message
  ----    ------             ----                   ----                   -------
  Normal  ScalingReplicaSet  31m                    deployment-controller  Scaled up replica set example-9-6ddd44564c to 2
  Normal  ScalingReplicaSet  3m56s                  deployment-controller  Scaled up replica set example-9-6ddd44564c to 2 from 1
  Normal  ScalingReplicaSet  3m56s                  deployment-controller  Scaled up replica set example-9-6755695d56 to 1
  Normal  ScalingReplicaSet  3m48s (x2 over 9m10s)  deployment-controller  Scaled down replica set example-9-6ddd44564c to 1 from 2
  Normal  ScalingReplicaSet  3m48s                  deployment-controller  Scaled up replica set example-9-6755695d56 to 2 from 1
  Normal  ScalingReplicaSet  3m34s                  deployment-controller  Scaled down replica set example-9-6ddd44564c to 0 from 1

Details of replicaset created by deployment

kubectl get rs

Output

controlplane $ kubectl get rs
NAME                   DESIRED   CURRENT   READY   AGE
example-9-6755695d56   2         2         2       47s
example-9-6ddd44564c   0         0         0       28m

Details of pods created by deployment in the replicaset

kubectl get pods

Output

controlplane $ kubectl get pods
NAME                         READY   STATUS    RESTARTS   AGE
example-9-6755695d56-vxkkq   1/1     Running   0          67s
example-9-6755695d56-xrm46   1/1     Running   0          59s

Verify the OS of the pod

kubectl exec -it example-9-6755695d56-vxkkq -- /bin/bash

Output

controlplane $ kubectl exec -it example-9-6755695d56-vxkkq -- cat /etc/os-release
NAME="CentOS Linux"
VERSION="8"
.
.
.

Now check for revision

kubectl rollout status deployment example-9
kubectl rollout history deployment example-9

Output

controlplane $ kubectl rollout status deployment example-9
Waiting for deployment "example-9" rollout to finish: 1 old replicas are pending termination...
Waiting for deployment "example-9" rollout to finish: 1 old replicas are pending termination...
deployment "example-9" successfully rolled out
controlplane $ kubectl rollout history deployment example-9
deployment.apps/example-9
REVISION  CHANGE-CAUSE
1         <none>
2         <none>

Now rollback to previous version and check for the pods OS version

kubectl rollout undo deployment example-9

Ouput

controlplane $ kubectl rollout undo deployment example-9
deployment.apps/example-9 rolled back

The current revision is 2 (can be verified from kubectl describe deployment example-9 command)

At the time of rollback the number of pods will be the number of desired state mentioned in manifest, even if you have scaled down (manually) in the previous version

Understanding Scalability

In very old versions of Kubernetes, the Deployment did not exist
To manage scalability, the ReplicaSet resource was used With the introduction of the apps API extension, the Deployment was introduced
From that moment, scalability was managed in the Deployment and no longer in the ReplicaSet
To manage scalability however, the Deployment creates a ReplicaSet
Do NOT manage ReplicaSets directly, manage through the Deployment only

Manually managing scalability

Use kubectl scale deployment my-deployment --repIicas=4 to scale the number of currently running replicas
Alternatively, use kubectl edit deployment my-deployment to edit the number of replicas manually -Notice that kubectl edit allows modification of a limited number of options only
To start a deployment with a specific number of replicas, use kubectl create deploy ... --replicas=3

Commands

Command	Description
`kubectl get deployment`	To Check deployment was created or not
`kubectl describe deployment example-9`	To check, how deployment create replica set & pods
`kubectl get rs`	List all replica sets in the current namespace.
`kubectl scale --replicas=1 deployment example-9`	To scale up or scale down
`kubectl logs -f POD_NAME`	View the logs of a specific pod.
`kubectl rollout status deployment example-9`	Check the status of the deployment's rollout process
`kubectl rollout history deployment example-9`	View the history of revisions for the specified deployment.
`kubectl rollout undo deployment example-9`	Roll back the deployment to the previous revision.
`kubectl rollout undo deployment example-9 --to-revision=2`	Roll back the deployment to the specific revision.

Replica Set Kubernetes Networking