Note_Tech

All technological notes.

Project maintained by simonangel-fong Hosted on GitHub Pages — Theme by mattgraham

Kubernetes: Deployment - Rollout & Rollback

Kubernetes: Deployment - Rollout \& Rollback

Imperative Command

Rollout

Command	Description
`kubectl rollout status deploy NAME`	Show the status of the rollout
`kubectl rollout history deploy NAME`	View rollout history
`kubectl rollout restart deploy NAME --selector=app=nginx`	Restart a resource with label
`kubectl rollout resume deploy NAME`	Resume a paused resource
`kubectl rollout pause deploy NAME`	Mark the provided resource as paused
`kubectl rollout undo deploy NAME`	Rollback to the previous version

Deployment Strategies

Deployment strategies
- help in defining how the new ReplicaSet should replace the existing ReplicaSet.

`Recreate`

kill all the existing RC and then bring up the new ones.
This results in quick deployment however it will result in downtime when the old pods are down and the new pods have not come up.
what happens underneath?
- when the pod template update
  1. Deployment controller update the desired replica to 0
  2. ReplicaSet Controller updates the old ReplicaSet’s replica to 0
  3. ReplicaSet Controller removes old Pod
  4. Deployment controller creates a new ReplicaSet with new hash
  5. ReplicaSet Controller create pod
the downtime occurts between the old and new rs

Lab: `Recreate` Strategy

# demo-strategy-recreate.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
  name: demo-strategy-recreate
spec:
  replicas: 4
  strategy:
    type: Recreate
  selector:
    matchLabels:
      app: demo-recreate
  template:
    metadata:
      labels:
        app: demo-recreate
    spec:
      containers:
        - image: web
          name: nginx

apply

kubectl apply -f demo-strategy-recreate.yaml
# deployment.apps/demo-strategy-recreate created

kubectl get deploy
# NAME                     READY   UP-TO-DATE   AVAILABLE   AGE
# demo-strategy-recreate   3/3     3            3           3m20s

update replica and apply
- pod won’t update because the template stay the same

kubectl apply -f demo-strategy-recreate.yaml
# deployment.apps/demo-strategy-recreate configured

kubectl get deploy
# NAME                     READY   UP-TO-DATE   AVAILABLE   AGE
# demo-strategy-recreate   4/4     4            4           5m30s

# confirm: check history
kubectl rollout history deployment/demo-strategy-recreate
# deployment.apps/demo-strategy-recreate
# REVISION  CHANGE-CAUSE
# 1         <none>

Update image
- image httpd

kubectl set image deploy demo-strategy-recreate web=httpd
# deployment.apps/demo-strategy-recreate image updated

# confirm: get update
kubectl rollout history deployment/demo-strategy-recreate
# deployment.apps/demo-strategy-recreate
# REVISION  CHANGE-CAUSE
# 1         <none>
# 2         <none>

# confirm:
#   Ready = 0 for a short time
kubectl get deploy -w
# NAME                     READY   UP-TO-DATE   AVAILABLE   AGE
# demo-strategy-recreate   4/4     4            4           79s
# demo-strategy-recreate   4/4     4            4           94s
# demo-strategy-recreate   4/4     0            4           94s
# demo-strategy-recreate   0/4     0            0           94s
# demo-strategy-recreate   0/4     0            0           97s
# demo-strategy-recreate   0/4     0            0           97s
# demo-strategy-recreate   0/4     4            0           97s
# demo-strategy-recreate   1/4     4            1           102s
# demo-strategy-recreate   2/4     4            2           104s
# demo-strategy-recreate   3/4     4            3           106s
# demo-strategy-recreate   4/4     4            4           106s

# confirm
kubectl rollout status

Update both image and labels

# linux
kubectl patch deploy demo-strategy-recreate --patch '
{
  "spec": {
    "template": {
      "metadata": {
        "labels": {
          "demo": "true"
        }
      },
      "spec": {
        "containers": [
          {
            "name": "web",
            "image": "traefik"
          }
        ]
      }
    }
  }
}'

# windows
kubectl patch deploy demo-strategy-recreate --patch "{\"spec\": {\"template\": {\"metadata\":{\"labels\":{\"demo\":\"true\"}}, \"spec\": {\"containers\": [{\"name\": \"web\", \"image\": \"traefik\"}]}}}}"
# deployment.apps/demo-strategy-recreate patched

# confirm: downtime exist
kubectl get deploy -o wide -w
# NAME                     READY   UP-TO-DATE   AVAILABLE   AGE   CONTAINERS   IMAGES   SELECTOR
# demo-strategy-recreate   4/4     4            4           15m   web          httpd    app=demo-recreate
# demo-strategy-recreate   4/4     4            4           19m   web          traefik   app=demo-recreate
# demo-strategy-recreate   4/4     0            4           19m   web          traefik   app=demo-recreate
# demo-strategy-recreate   0/4     0            0           19m   web          traefik   app=demo-recreate
# demo-strategy-recreate   0/4     0            0           19m   web          traefik   app=demo-recreate
# demo-strategy-recreate   0/4     0            0           19m   web          traefik   app=demo-recreate
# demo-strategy-recreate   0/4     4            0           19m   web          traefik   app=demo-recreate
# demo-strategy-recreate   1/4     4            1           19m   web          traefik   app=demo-recreate
# demo-strategy-recreate   2/4     4            2           19m   web          traefik   app=demo-recreate
# demo-strategy-recreate   3/4     4            3           19m   web          traefik   app=demo-recreate
# demo-strategy-recreate   4/4     4            4           19m   web          traefik   app=demo-recreate

# confirm: update
kubectl rollout history deployment/demo-strategy-recreate
# deployment.apps/demo-strategy-recreate
# REVISION  CHANGE-CAUSE
# 1         <none>
# 2         <none>
# 3         <none>

# confirm the underlying rs
kubectl get rs
# NAME                                DESIRED   CURRENT   READY   AGE
# demo-strategy-recreate-68cd744854   4         4         4       4m52s
# demo-strategy-recreate-6f979ddc4f   0         0         0       24m
# demo-strategy-recreate-764d5b8785   0         0         0       22m

`Rolling Update`

default
gradually brings down the old RS and brings up the new one.
This results in slow deployment, however there is no deployment.
At all times, few old pods and few new pods are available in this process.
the Pods are replaced gradually, by simultaneously
- scaling down the old ReplicaSet
- scaling up the new ReplicaSet
what happens underneath?
- when the pod template update
  1. Deployment Controller creates a new ReplicaSet with new hash
  2. Deployment Controller scale down the old ReplicaSet and scale up the new ReplicaSet gradually, while keeping the total pod number close to the desired number.
  3. ReplicaSet Controller creates Pod in new ReplicaSet and removes Pod in old ReplicaSet
  4. Repeat process
  5. Until old ReplicaSet scale down to 0 and new ReplicaSet scale up to desired number.

Parameters

spec.strategy.rollingUpdate.maxSurge:
- The maximum number of Pods above the desired number of replicas that the Deployment can have during the rolling update.
- can be
  - an absolute number
  - a percentage of the desired number
- default: 25%

spec.strategy.rollingUpdate.maxUnavailable:
- The maximum number of Pods relative to the desired replica count that can be unavailable during the rolling update.
- can be
  - an absolute number
  - a percentage of the desired number
- 25%

example:
- MaxSurge=0; maxUnavailable=0:
  - unavailable
  - no pod be replace at any time
- replicas=3; MaxSurge=0; maxUnavailable=1:
  - total pod always = 3+0 = 3
  - avaible always = 3-1 = 2
  - scale down the old rs first, then scale up the new rs
- replicas=3; MaxSurge=1; maxUnavailable=0:
  - total pod always = 3+1 = 4
  - avaible always = 3-0 = 3
  - scale up the new rs first, then scale down the old rs
- replicas=3; maxSurge=1, maxUnavailable=1:
  - total pod always = 3+1 = 4
  - avaible always = 3-1 = 2
  - scale down the old rs and scale up the new rs; when one new pod available, one old pod get removed.
- replicas=10; maxSurge=25%, maxUnavailable=25%:
  - total pod = 10 * 125% ~= 13
  - available pod = 10 * 75% ~= 8
- higher values of maxSurge and maxUnavailable
  - higher maxSurge: more new pod created in the new rs
  - higher maxUnavailable: more old pod get removed in the old rs

Lab: `RollingUdate` Strategy

# demo-strategy-rollingupdate.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
  name: demo-strategy-rollingupdate
spec:
  replicas: 3
  strategy:
    type: RollingUpdate
    rollingUpdate:
      maxSurge: 0
      maxUnavailable: 1
  minReadySeconds: 10
  selector:
    matchLabels:
      app: demo-rollingupdate
  template:
    metadata:
      labels:
        app: demo-rollingupdate
    spec:
      containers:
        - image: web
          name: nginx

kubectl apply -f demo-strategy-rollingupdate.yaml
# deployment.apps/demo-strategy-rollingupdate created

# get the rollout status
kubectl rollout status deploy demo-strategy-rollingupdate
# Waiting for deployment "demo-strategy-rollingupdate" rollout to finish: 0 of 3 updated replicas are available...
# Waiting for deployment "demo-strategy-rollingupdate" rollout to finish: 0 of 3 updated replicas are available...
# Waiting for deployment "demo-strategy-rollingupdate" rollout to finish: 0 of 3 updated replicas are available...
# Waiting for deployment "demo-strategy-rollingupdate" rollout to finish: 1 of 3 updated replicas are available...
# Waiting for deployment "demo-strategy-rollingupdate" rollout to finish: 2 of 3 updated replicas are available...
# deployment "demo-strategy-rollingupdate" successfully rolled out

# confirm rs by monitoring:
kubectl get rs -w
# NAME                                     DESIRED   CURRENT   READY   AGE
# demo-strategy-rollingupdate-79f49b4c46   3         0         0       0s
# demo-strategy-rollingupdate-79f49b4c46   3         0         0       0s
# demo-strategy-rollingupdate-79f49b4c46   3         3         0       0s
# demo-strategy-rollingupdate-79f49b4c46   3         3         1       3s
# demo-strategy-rollingupdate-79f49b4c46   3         3         2       4s
# demo-strategy-rollingupdate-79f49b4c46   3         3         3       8s
# demo-strategy-rollingupdate-79f49b4c46   3         3         3       13s
# demo-strategy-rollingupdate-79f49b4c46   3         3         3       14s
# demo-strategy-rollingupdate-79f49b4c46   3         3         3       18s

# confirm deploy by monitoring:
kubectl get deploy -w
# NAME                          READY   UP-TO-DATE   AVAILABLE   AGE
# demo-strategy-rollingupdate   0/3     0            0           0s
# demo-strategy-rollingupdate   0/3     0            0           0s
# demo-strategy-rollingupdate   0/3     0            0           0s
# demo-strategy-rollingupdate   0/3     3            0           0s
# demo-strategy-rollingupdate   1/3     3            0           3s
# demo-strategy-rollingupdate   2/3     3            0           4s
# demo-strategy-rollingupdate   3/3     3            0           8s
# demo-strategy-rollingupdate   3/3     3            1           13s
# demo-strategy-rollingupdate   3/3     3            2           14s
# demo-strategy-rollingupdate   3/3     3            3           18s

# confirm history
kubectl rollout history deploy demo-strategy-rollingupdate
# deployment.apps/demo-strategy-rollingupdate
# REVISION  CHANGE-CAUSE
# 1         <none>

Set image

kubectl set image deploy demo-strategy-rollingupdate web=httpd
# deployment.apps/demo-strategy-rollingupdate image updated

# confirm rs by monitoring:
# old rs scale down and new rs scale up gradually
kubectl get rs -w
# NAME                                     DESIRED   CURRENT   READY   AGE
# demo-strategy-rollingupdate-79f49b4c46   3         3         3       2m26s
# demo-strategy-rollingupdate-6b9cc477f4   0         0         0       0s
# demo-strategy-rollingupdate-6b9cc477f4   0         0         0       0s
# demo-strategy-rollingupdate-79f49b4c46   2         3         3       2m41s
# demo-strategy-rollingupdate-6b9cc477f4   1         0         0       0s
# demo-strategy-rollingupdate-79f49b4c46   2         3         3       2m41s
# demo-strategy-rollingupdate-79f49b4c46   2         2         2       2m41s
# demo-strategy-rollingupdate-6b9cc477f4   1         0         0       0s
# demo-strategy-rollingupdate-6b9cc477f4   1         1         0       0s
# demo-strategy-rollingupdate-6b9cc477f4   1         1         1       4s
# demo-strategy-rollingupdate-6b9cc477f4   1         1         1       16s
# demo-strategy-rollingupdate-79f49b4c46   1         2         2       2m57s
# demo-strategy-rollingupdate-79f49b4c46   1         2         2       2m57s
# demo-strategy-rollingupdate-6b9cc477f4   2         1         1       16s
# demo-strategy-rollingupdate-79f49b4c46   1         1         1       2m57s
# demo-strategy-rollingupdate-6b9cc477f4   2         1         1       16s
# demo-strategy-rollingupdate-6b9cc477f4   2         2         1       16s
# demo-strategy-rollingupdate-6b9cc477f4   2         2         2       18s
# demo-strategy-rollingupdate-6b9cc477f4   2         2         2       28s
# demo-strategy-rollingupdate-79f49b4c46   0         1         1       3m9s
# demo-strategy-rollingupdate-79f49b4c46   0         1         1       3m9s
# demo-strategy-rollingupdate-6b9cc477f4   3         2         2       28s
# demo-strategy-rollingupdate-6b9cc477f4   2         2         1       16s
# demo-strategy-rollingupdate-79f49b4c46   0         0         0       3m9s
# demo-strategy-rollingupdate-6b9cc477f4   3         2         2       28s
# demo-strategy-rollingupdate-6b9cc477f4   3         3         2       28s
# demo-strategy-rollingupdate-6b9cc477f4   3         3         3       30s
# demo-strategy-rollingupdate-6b9cc477f4   3         3         3       41s

# confirm deploy by monitoring:
#  always ready: 2
kubectl get deploy -w
# NAME                          READY   UP-TO-DATE   AVAILABLE   AGE
# demo-strategy-rollingupdate   3/3     3            3           2m34s
# demo-strategy-rollingupdate   3/3     3            3           2m41s
# demo-strategy-rollingupdate   3/3     3            3           2m41s
# demo-strategy-rollingupdate   3/3     0            3           2m41s
# demo-strategy-rollingupdate   2/3     0            2           2m41s
# demo-strategy-rollingupdate   2/3     1            2           2m41s
# demo-strategy-rollingupdate   3/3     1            2           2m45s
# demo-strategy-rollingupdate   3/3     1            3           2m57s
# demo-strategy-rollingupdate   2/3     1            2           2m57s
# demo-strategy-rollingupdate   2/3     2            2           2m57s
# demo-strategy-rollingupdate   3/3     2            2           2m59s
# demo-strategy-rollingupdate   2/3     1            2           2m41s
# demo-strategy-rollingupdate   3/3     2            3           3m9s
# demo-strategy-rollingupdate   2/3     2            2           3m9s
# demo-strategy-rollingupdate   2/3     3            2           3m9s
# demo-strategy-rollingupdate   3/3     3            2           3m11s
# demo-strategy-rollingupdate   3/3     3            3           3m22s

Control Rollout Process

spec.progressDeadlineSeconds field
- configure the rollout progress deadline
  - rollout process never stops completely when reaching deadline
- default: 600s
- confirm by status.conditions.Progressing

Pausing and Resuming the Rollout Process

kubectl rollout pause deployment kiada
- sets spec.paused = true
- Deployment controller checks this field before any change to the underlying ReplicaSets.
deployment running both the old and new versions
kubectl rollout resume deployment kiada

Preventing a Faulty Version

Ready vs Available
- Ready:
  - the pods are passing readiness probes
- Available:
  - the pod is ready for a specific amount of time
  - the rolling update continues only if the new pod is avaialbe.
minReadySeconds field
- used to define the amount of time between ready and avaialbe.
- default: 0
  - the pods are available when reaady.
- during this time, a Pod can receives client requests.

With `Recreate` Strategy

Scenario:
- replicas: 3
- strategy: Recreate
- minReadySeconds: 60
How it works:
- new manifest applys
- Deployment Controller first scales down the desired replica of the old ReplicaSet to zero;
- ReplicaSet Controller terminates the old pods of the old ReplicaSet
- Deployment Controller waits for the Pods to be fully terminated;
- Deployment Controller creates a new ReplicaSet and scale up
- ReplicaSet Controller creates new pod aligned with the new ReplicaSet
- New pods pass Reaadiness Probe; Pods accept traffic.
- The Deployment
  - shows status of READY 3/3 but AVAILABLE 0
  - waits until the minReadySeconds: 60
- If pods stay healthy after minReadySeconds, deployment status shows AVAILABLE 3/3

Phase	Deployment Status	Real-World Impact
Start	`3/3 READY, 3 AVAILABLE`	Old version running.
Scale Down	`0/0 READY, 0 AVAILABLE`	Downtime begins. All old pods are killed.
Scale Up	`0/3 READY, 0 AVAILABLE`	New pods are starting (ContainerCreating).
Probes Pass	`3/3 READY, 0 AVAILABLE`	Downtime ends. Traffic flows to new pods.
Probation	`3/3 READY, 0 AVAILABLE`	minReadySeconds timer is counting
Success	`3/3 READY, 3 AVAILABLE`	Rollout complet

Lab: Recreate with `minReadySeconds`

# demo-recreate-minreadyseconds.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
  name: demo-recreate-minreadyseconds
spec:
  replicas: 3
  minReadySeconds: 60
  strategy:
    type: Recreate
  selector:
    matchLabels:
      app: web
  template:
    metadata:
      labels:
        app: web
    spec:
      containers:
        - image: nginx
          name: web

kubectl apply -f demo-recreate-minreadyseconds.yaml
# deployment.apps/demo-recreate-minreadyseconds created

# update image
kubectl set image deploy demo-recreate-minreadyseconds web=httpd
# deployment.apps/demo-recreate-minreadyseconds image updated


# rollout status: wait until 60s
kubectl rollout status deploy demo-recreate-minreadyseconds
# Waiting for deployment "demo-recreate-minreadyseconds" rollout to finish: 0 of 3 updated replicas are available...
# Waiting for deployment "demo-recreate-minreadyseconds" rollout to finish: 0 of 3 updated replicas are available...
# Waiting for deployment "demo-recreate-minreadyseconds" rollout to finish: 0 of 3 updated replicas are available...
# Waiting for deployment "demo-recreate-minreadyseconds" rollout to finish: 0 of 3 updated replicas are available...
# deployment "demo-recreate-minreadyseconds" successfully rolled out

# deploy status:
#   READY=0 for short time
#   all ready: 4m36s
#   all available: 5m36s
kubectl get deploy -w
# NAME                            READY   UP-TO-DATE   AVAILABLE   AGE
# demo-recreate-minreadyseconds   3/3     3            3           3m50s
# demo-recreate-minreadyseconds   3/3     3            3           4m28s
# demo-recreate-minreadyseconds   3/3     0            3           4m28s
# demo-recreate-minreadyseconds   0/3     0            0           4m28s
# demo-recreate-minreadyseconds   0/3     0            0           4m30s
# demo-recreate-minreadyseconds   0/3     0            0           4m30s
# demo-recreate-minreadyseconds   0/3     3            0           4m30s
# demo-recreate-minreadyseconds   1/3     3            0           4m34s
# demo-recreate-minreadyseconds   2/3     3            0           4m34s
# demo-recreate-minreadyseconds   3/3     3            0           4m36s
# demo-recreate-minreadyseconds   3/3     3            3           5m36s

# rs:
#  old rs scale down to 0
#  create new rs
#  new rs scale up
kubectl get rs -w
# NAME                                       DESIRED   CURRENT   READY   AGE
# demo-recreate-minreadyseconds-5f876cbdc4   3         3         3       3m57s
# demo-recreate-minreadyseconds-5f876cbdc4   0         3         3       4m28s
# demo-recreate-minreadyseconds-5f876cbdc4   0         3         3       4m28s
# demo-recreate-minreadyseconds-5f876cbdc4   0         0         0       4m28s
# demo-recreate-minreadyseconds-59778758cd   3         0         0       0s
# demo-recreate-minreadyseconds-59778758cd   3         0         0       1s
# demo-recreate-minreadyseconds-59778758cd   3         3         0       1s
# demo-recreate-minreadyseconds-59778758cd   3         3         1       5s
# demo-recreate-minreadyseconds-59778758cd   3         3         2       5s
# demo-recreate-minreadyseconds-59778758cd   3         3         3       7s
# demo-recreate-minreadyseconds-59778758cd   3         3         3       67s

With RollingOut Strategy

Scenario:
- replicas: 3
- strategy: RollingUpdate
- maxSurge: 1
- maxUnavailable: 0
- minReadySeconds: 60
How it works:
- New manifest applies.
- Deployment Controller creates a new ReplicaSet (New RS).
- Deployment Controller scales the new ReplicaSet to 1 (calculated as replicas + maxSurge).
- ReplicaSet Controller creates the first new pod.
- New pod passes Readiness Probe; Pod begins accepting traffic.
- The Deployment shows READY 4/3 but AVAILABLE 3.
- Deployment Controller waits until the minReadySeconds: 60 timer finishes for that specific pod.
- Once the pod is Available, the Deployment Controller scales down the old ReplicaSet to 2.
- The cycle repeats:
  - It “surges” a second new pod, waits 60s, then kills a second old pod.
- This continues until only the 3 new pods remain.

Phase	Pod Count (Old/New)	Deployment Status	Real-World Impact
Start	3 Old / 0 New	`3/3 READY, 3 AVAILABLE`	Only old version running.
First Surge	3 Old / 1 New	`4/3 READY, 3 AVAILABLE`	New pod starting; total 4 pods.
First Ready	3 Old / 1 New	`4/3 READY, 3 AVAILABLE`	Traffic hits new pod. Timer starts.
First Available	3 Old / 1 New	`4/3 READY, 4 AVAILABLE`	New pod passed 60s probation.
Scale Down	2 Old / 1 New	`3/3 READY, 3 AVAILABLE`	One old pod killed. Total back to 3.
Second Surge	2 Old / 2 New	`4/3 READY, 3 AVAILABLE`	Process repeats for the next pod.
Success	0 Old / 3 New	`3/3 READY, 3 AVAILABLE`	Rollout complete; 0 downtime.

Lab: RollingOut with `minReadySeconds`

# demo-rollingout-minreadyseconds.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
  name: demo-rollingout-minreadyseconds
spec:
  replicas: 3
  strategy:
    type: RollingUpdate
    rollingUpdate:
      maxSurge: 1
      maxUnavailable: 0
  minReadySeconds: 60
  selector:
    matchLabels:
      app: web
  template:
    metadata:
      labels:
        app: web
    spec:
      containers:
        - name: web
          image: nginx

kubectl apply -f demo-rollingout-minreadyseconds.yaml
# deployment.apps/demo-rollingout-minreadyseconds created

# update image
kubectl set image deploy demo-rollingout-minreadyseconds web=httpd
# deployment.apps/demo-rollingout-minreadyseconds image updated


# rollout status: wait until 60s
kubectl rollout status deploy demo-rollingout-minreadyseconds
# Waiting for deployment "demo-rollingout-minreadyseconds" rollout to finish: 1 out of 3 new replicas have been updated...
# Waiting for deployment "demo-rollingout-minreadyseconds" rollout to finish: 1 out of 3 new replicas have been updated...
# Waiting for deployment "demo-rollingout-minreadyseconds" rollout to finish: 1 out of 3 new replicas have been updated...
# Waiting for deployment "demo-rollingout-minreadyseconds" rollout to finish: 1 out of 3 new replicas have been updated...
# Waiting for deployment "demo-rollingout-minreadyseconds" rollout to finish: 2 out of 3 new replicas have been updated...
# Waiting for deployment "demo-rollingout-minreadyseconds" rollout to finish: 2 out of 3 new replicas have been updated...
# Waiting for deployment "demo-rollingout-minreadyseconds" rollout to finish: 2 out of 3 new replicas have been updated...
# Waiting for deployment "demo-rollingout-minreadyseconds" rollout to finish: 2 out of 3 new replicas have been updated...
# Waiting for deployment "demo-rollingout-minreadyseconds" rollout to finish: 1 old replicas are pending termination...
# Waiting for deployment "demo-rollingout-minreadyseconds" rollout to finish: 1 old replicas are pending termination...
# Waiting for deployment "demo-rollingout-minreadyseconds" rollout to finish: 1 old replicas are pending termination...
# deployment "demo-rollingout-minreadyseconds" successfully rolled out

# deploy status:
# repeat process:
#   READY & avaialable: up to 4 and down to 4
kubectl get deploy -w
# NAME                              READY   UP-TO-DATE   AVAILABLE   AGE
# demo-rollingout-minreadyseconds   3/3     3            0           22s
# demo-rollingout-minreadyseconds   3/3     3            3           68s
# demo-rollingout-minreadyseconds   3/3     3            3           81s
# demo-rollingout-minreadyseconds   3/3     3            3           81s
# demo-rollingout-minreadyseconds   3/3     0            3           81s
# demo-rollingout-minreadyseconds   3/3     1            3           81s
# demo-rollingout-minreadyseconds   4/3     1            3           84s
# demo-rollingout-minreadyseconds   4/3     1            4           2m26s
# demo-rollingout-minreadyseconds   3/3     1            3           2m26s
# demo-rollingout-minreadyseconds   3/3     2            3           2m26s
# demo-rollingout-minreadyseconds   4/3     2            3           2m29s
# demo-rollingout-minreadyseconds   4/3     2            4           3m31s
# demo-rollingout-minreadyseconds   3/3     2            3           3m31s
# demo-rollingout-minreadyseconds   3/3     3            3           3m31s
# demo-rollingout-minreadyseconds   4/3     3            3           3m34s
# demo-rollingout-minreadyseconds   4/3     3            4           4m36s
# demo-rollingout-minreadyseconds   3/3     3            3           4m36s

# rs:
kubectl get rs -w
# NAME                                         DESIRED   CURRENT   READY   AGE
# demo-rollingout-minreadyseconds-5f876cbdc4   3         3         3       27s
# demo-rollingout-minreadyseconds-5f876cbdc4   3         3         3       68s
# demo-rollingout-minreadyseconds-59778758cd   1         0         0       0s
# demo-rollingout-minreadyseconds-59778758cd   1         0         0       0s
# demo-rollingout-minreadyseconds-59778758cd   1         1         0       0s
# demo-rollingout-minreadyseconds-59778758cd   1         1         1       3s
# demo-rollingout-minreadyseconds-59778758cd   1         1         1       65s
# demo-rollingout-minreadyseconds-5f876cbdc4   2         3         3       2m26s
# demo-rollingout-minreadyseconds-59778758cd   2         1         1       65s
# demo-rollingout-minreadyseconds-5f876cbdc4   2         3         3       2m26s
# demo-rollingout-minreadyseconds-59778758cd   2         1         1       65s
# demo-rollingout-minreadyseconds-5f876cbdc4   2         2         2       2m26s
# demo-rollingout-minreadyseconds-59778758cd   2         2         1       65s
# demo-rollingout-minreadyseconds-59778758cd   2         2         2       68s
# demo-rollingout-minreadyseconds-59778758cd   2         2         2       2m10s
# demo-rollingout-minreadyseconds-5f876cbdc4   1         2         2       3m31s
# demo-rollingout-minreadyseconds-59778758cd   3         2         2       2m10s
# demo-rollingout-minreadyseconds-5f876cbdc4   1         2         2       3m31s
# demo-rollingout-minreadyseconds-5f876cbdc4   1         1         1       3m31s
# demo-rollingout-minreadyseconds-59778758cd   3         2         2       2m10s
# demo-rollingout-minreadyseconds-59778758cd   3         3         2       2m10s
# demo-rollingout-minreadyseconds-59778758cd   3         3         3       2m13s
# demo-rollingout-minreadyseconds-59778758cd   3         3         3       3m15s
# demo-rollingout-minreadyseconds-5f876cbdc4   0         1         1       4m36s
# demo-rollingout-minreadyseconds-5f876cbdc4   0         1         1       4m36s
# demo-rollingout-minreadyseconds-5f876cbdc4   0         0         0       4m36s

Lab: RollingOut with `minReadySeconds` if a faulty version

# demo-rollingout-minreadyseconds-faulty.yaml
apiVersion: apps/v1
kind: Deployment
metadata:
  name: demo-rollingout-minreadyseconds-faulty
spec:
  replicas: 3
  strategy:
    type: RollingUpdate
    rollingUpdate:
      maxSurge: 1
      maxUnavailable: 0
  minReadySeconds: 60
  selector:
    matchLabels:
      app: web
  template:
    metadata:
      labels:
        app: web
    spec:
      containers:
        - name: web
          image: nginx

kubectl apply -f demo-rollingout-minreadyseconds-faulty.yaml
# deployment.apps/demo-rollingout-minreadyseconds-faulty created

# update faulty image
kubectl set image deploy demo-rollingout-minreadyseconds-faulty web=web

# confirm: rollout get stuck
kubectl rollout status deploy demo-rollingout-minreadyseconds-faulty
# Waiting for deployment "demo-rollingout-minreadyseconds-faulty" rollout to finish: 1 out of 3 new replicas have been updated...
# Waiting for deployment "demo-rollingout-minreadyseconds-faulty" rollout to finish: 1 out of 3 new replicas have been updated...
# error: deployment "demo-rollingout-minreadyseconds-faulty" exceeded its progress deadline

# deploy status:
kubectl get deploy -w
# NAME                                     READY   UP-TO-DATE   AVAILABLE   AGE
# demo-rollingout-minreadyseconds-faulty   3/3     3            3           2m58s
# demo-rollingout-minreadyseconds-faulty   3/3     3            3           3m19s
# demo-rollingout-minreadyseconds-faulty   3/3     3            3           3m19s
# demo-rollingout-minreadyseconds-faulty   3/3     0            3           3m19s
# demo-rollingout-minreadyseconds-faulty   3/3     1            3           3m19s

# rs:
kubectl get rs -w
# NAME                                                DESIRED   CURRENT   READY   AGE
# demo-rollingout-minreadyseconds-faulty-5f876cbdc4   3         3         3       3m6s
# demo-rollingout-minreadyseconds-faulty-7fc574699b   1         0         0       0s
# demo-rollingout-minreadyseconds-faulty-7fc574699b   1         0         0       0s
# demo-rollingout-minreadyseconds-faulty-7fc574699b   1         1         0       0s

kubectl get pod
# NAME                                                      READY   STATUS             RESTARTS   AGE
# demo-rollingout-minreadyseconds-faulty-5f876cbdc4-9zgdr   1/1     Running            0          5m23s
# demo-rollingout-minreadyseconds-faulty-5f876cbdc4-p4sd4   1/1     Running            0          5m23s
# demo-rollingout-minreadyseconds-faulty-5f876cbdc4-w7xtk   1/1     Running            0          5m23s
# demo-rollingout-minreadyseconds-faulty-7fc574699b-rqqgb   0/1     ImagePullBackOff   0          2m4s

rolling update gets stuck because the image is incorrect and the maxUnavailable: 0, which does not allow unavailable pod. therefore deployment rolling update provide a safety net.

fix: rollback

# check history before rollback
kubectl rollout history deployment demo-rollingout-minreadyseconds-faulty
# deployment.apps/demo-rollingout-minreadyseconds-faulty
# REVISION  CHANGE-CAUSE
# 1         <none>
# 2         <none>

# rollback deploy
kubectl rollout undo deployment demo-rollingout-minreadyseconds-faulty
# deployment.apps/demo-rollingout-minreadyseconds-faulty rolled back

# deploy status
kubectl get deploy -w
# NAME                                     READY   UP-TO-DATE   AVAILABLE   AGE
# demo-rollingout-minreadyseconds-faulty   3/3     1            3           17m
# demo-rollingout-minreadyseconds-faulty   3/3     1            3           18m
# demo-rollingout-minreadyseconds-faulty   3/3     1            3           18m
# demo-rollingout-minreadyseconds-faulty   3/3     3            3           18m
# demo-rollingout-minreadyseconds-faulty   3/3     3            3           18m

kubectl get rs -w
# NAME                                                DESIRED   CURRENT   READY   AGE
# demo-rollingout-minreadyseconds-faulty-5f876cbdc4   3         3         3       17m
# demo-rollingout-minreadyseconds-faulty-7fc574699b   1         1         0       13m
# demo-rollingout-minreadyseconds-faulty-5f876cbdc4   3         3         3       18m
# demo-rollingout-minreadyseconds-faulty-7fc574699b   0         1         0       14m
# demo-rollingout-minreadyseconds-faulty-7fc574699b   0         1         0       14m
# demo-rollingout-minreadyseconds-faulty-7fc574699b   0         0         0       14m

kubectl rollout history deployment demo-rollingout-minreadyseconds-faulty
# deployment.apps/demo-rollingout-minreadyseconds-faulty
# REVISION  CHANGE-CAUSE
# 2         <none>
# 3         <none>

# confirm: image roll back
kubectl get deploy -o wide
# NAME                                     READY   UP-TO-DATE   AVAILABLE   AGE   CONTAINERS   IMAGES   SELECTOR
# demo-rollingout-minreadyseconds-faulty   3/3     3            3           20m   web          nginx    app=web

Rollout History

CHANGE-CAUSE colume
- pass message by the --record option of the kubectl command
  - has deprecated
inspect each revision
- kubectl rollout history deploy NAME --revision 2
revision history
- represented by the ReplicaSets associated with the Deployment
deploy.spec.revisionHistoryLimit
- specify the size of the revision history, and thus the number of ReplicaSets that the Deployment controller keeps for a given Deployment,
- default: 10
revision numbe also stored in the ReplicaSet’s annotations

Rollout Restart

deletes and replaces the Pods using the same strategy used for updates.
refresh Pods without changing anything in the configuration.
it follows the deployment strategy
- RollingUpdate strategy: the Pods are recreated gradually
- Recreate: deleted and recreated simultaneously.

Rollback

Rollback
- undo a change
- Behind the scence:
  - Rollback creates a new rs, then take one pod from the new version rs, and bring up one pod in old rs, until all pods in new rs are down.
  - Can be verify duration deployment upgrade, kubectl get rs
if the RollingUpdate strategy is used, the Pods are rolled back gradually.
The kubectl rollout undo command can be used while the rollout process is running to cancel the rollout, or after the rollout is complete to undo it.
When a Deployment is paused with the kubectl pause command, the kubectl rollout undo command does nothing until you resume the Deployment with kubectl rollout resume.

Rolling Back vs Apply Old Manifest

kubectl rollout undo command:
- reverts only the Pod template
- preserves any other changes you made to the Deployment manifest.
  - e.g.,
    - update strategy
    - desired number of replicas
kubectl apply command:
- overwrites these changes that no included in the manifest.

Note_Tech

Kubernetes: Deployment - Rollout & Rollback

Imperative Command

Deployment Strategies

Recreate

Lab: Recreate Strategy

Rolling Update

Parameters

Lab: RollingUdate Strategy

Control Rollout Process

Pausing and Resuming the Rollout Process

Preventing a Faulty Version

With Recreate Strategy

Lab: Recreate with minReadySeconds

With RollingOut Strategy

Lab: RollingOut with minReadySeconds

Lab: RollingOut with minReadySeconds if a faulty version

Rollout History

Rollout Restart

Rollback

Rolling Back vs Apply Old Manifest

`Recreate`

Lab: `Recreate` Strategy

`Rolling Update`

Lab: `RollingUdate` Strategy

With `Recreate` Strategy

Lab: Recreate with `minReadySeconds`

Lab: RollingOut with `minReadySeconds`

Lab: RollingOut with `minReadySeconds` if a faulty version