Deployment — ReplicaSet, rolling update আর rollback

apiVersion: apps/v1
kind: Deployment
metadata:
  name: web
  labels:
    app: web
spec:
  replicas: 4
  revisionHistoryLimit: 10
  progressDeadlineSeconds: 40
  selector:
    matchLabels:
      app: web
  strategy:
    type: RollingUpdate
    rollingUpdate:
      maxSurge: 1
      maxUnavailable: 1
  template:
    metadata:
      labels:
        app: web
    spec:
      terminationGracePeriodSeconds: 30
      containers:
        - name: nginx
          image: nginx:1.27-alpine
          imagePullPolicy: IfNotPresent
          ports:
            - containerPort: 80
          readinessProbe:
            httpGet:
              path: /
              port: 80
            initialDelaySeconds: 2
            periodSeconds: 2
          resources:
            requests:
              cpu: 20m
              memory: 16Mi
            limits:
              cpu: 200m
              memory: 64Mi

$ kubectl apply -f web.yaml
deployment.apps/web created

$ kubectl rollout status deploy/web
Waiting for deployment "web" rollout to finish: 0 of 4 updated replicas are available...
Waiting for deployment "web" rollout to finish: 1 of 4 updated replicas are available...
Waiting for deployment "web" rollout to finish: 2 of 4 updated replicas are available...
Waiting for deployment "web" rollout to finish: 3 of 4 updated replicas are available...
deployment "web" successfully rolled out

$ kubectl get deploy web
NAME   READY   UP-TO-DATE   AVAILABLE   AGE
web    4/4     4            4           3s

$ kubectl get rs
NAME            DESIRED   CURRENT   READY   AGE
web-df7bd6557   4         4         4       3s

$ kubectl get pods
NAME                  READY   STATUS    RESTARTS   AGE
web-df7bd6557-cbs4s   1/1     Running   0          3s
web-df7bd6557-fq29n   1/1     Running   0          3s
web-df7bd6557-m49rk   1/1     Running   0          3s
web-df7bd6557-snjmn   1/1     Running   0          3s
…

# kubectl get pod web-df7bd6557-cbs4s -o yaml (ছাঁটা)
  ownerReferences:
  - apiVersion: apps/v1
    blockOwnerDeletion: true
    controller: true
    kind: ReplicaSet
    name: web-df7bd6557
    uid: ae844ce6-0df6-4324-b799-04c2fbaaeffd

# kubectl get rs web-df7bd6557 -o yaml (ছাঁটা)
  ownerReferences:
  - apiVersion: apps/v1
    blockOwnerDeletion: true
    controller: true
    kind: Deployment
    name: web
    uid: ce65b353-c8e2-4682-85b2-fd1e4dbb1562

$ kubectl scale deploy/web --replicas=6
deployment.apps/web scaled

$ kubectl get rs
NAME            DESIRED   CURRENT   READY   AGE
web-df7bd6557   6         6         6       9s

$ kubectl scale deploy/web --replicas=4
deployment.apps/web scaled

$ kubectl get rs
NAME            DESIRED   CURRENT   READY   AGE
web-df7bd6557   4         4         4       12s

$ kubectl rollout history deploy/web
deployment.apps/web
REVISION  CHANGE-CAUSE
1         <none>

$ kubectl patch deploy web -p '{"spec":{"selector":{"matchLabels":{"app":"shop"}}}}'
The Deployment "web" is invalid:
* spec.template.metadata.labels: Invalid value: {"app":"web"}: `selector` does not match template `labels`
* spec.selector: Invalid value: {"matchLabels":{"app":"shop"}}: field is immutable

$ kubectl label pod web-59d9464878-75t94 app=debug --overwrite
pod/web-59d9464878-75t94 labeled

$ kubectl get pods -L app
NAME                   READY   STATUS    RESTARTS   AGE     APP
web-59d9464878-75t94   1/1     Running   0          41s     debug
web-59d9464878-g49ql   1/1     Running   0          41s     web
web-59d9464878-hkcv8   1/1     Running   0          4s      web
web-59d9464878-hpf28   1/1     Running   0          41s     web
web-59d9464878-krxwg   1/1     Running   0          41s     web
…
# ওই Pod-এর ownerReferences এখন খালি: owners=[]

// How maxSurge and maxUnavailable percentages become Pod counts.
// maxSurge rounds UP, maxUnavailable rounds DOWN.
const surgePct = 25, unavailPct = 25;   // the Kubernetes defaults; try 10 and 10

function resolve(R) {
  let surge = Math.ceil(R * surgePct / 100);
  let unavail = Math.floor(R * unavailPct / 100);
  if (surge === 0 && unavail === 0) unavail = 1;   // both zero could never move
  return { R, surge, unavail, maxTotal: R + surge, minAvail: R - unavail };
}

console.log("replicas surge unavail maxTotal minAvailable");
for (const R of [1, 2, 3, 4, 5, 8, 10, 20, 100]) {
  const r = resolve(R);
  console.log([r.R, r.surge, r.unavail, r.maxTotal, r.minAvail]
    .map(v => String(v).padStart(6)).join(" "));
}

$ kubectl set image deploy/web nginx=nginx:1.28-alpine
deployment.apps/web image updated

$ kubectl rollout status deploy/web
Waiting for deployment "web" rollout to finish: 2 out of 4 new replicas have been updated...
…
Waiting for deployment "web" rollout to finish: 3 out of 4 new replicas have been updated...
…
Waiting for deployment "web" rollout to finish: 1 old replicas are pending termination...
…
Waiting for deployment "web" rollout to finish: 3 of 4 updated replicas are available...
deployment "web" successfully rolled out

$ kubectl get rs
NAME             DESIRED   CURRENT   READY   AGE
web-84c4bfd5dd   4         4         4       8s
web-df7bd6557    0         0         0       32s

# kubectl describe deploy web-এর event থেকে:
Scaled up replica set web-84c4bfd5dd from 0 to 1
Scaled down replica set web-df7bd6557 from 4 to 3
Scaled up replica set web-84c4bfd5dd from 1 to 2
Scaled down replica set web-df7bd6557 from 3 to 2
Scaled up replica set web-84c4bfd5dd from 2 to 3
Scaled down replica set web-df7bd6557 from 2 to 1
…

// Rolling update, following the Deployment controller's two rules.
// Change these lines and press Run again.
const R = 4;              // spec.replicas
const surge = 1;          // maxSurge, as a Pod count
const unavail = 1;        // maxUnavailable, as a Pod count
const readyAfter = 3;     // seconds from Pod creation to Ready

const maxTotal = R + surge, minAvail = R - unavail;
let old = R, clock = 0, newPods = [];    // newPods = creation times
const ready = () => newPods.filter(t => clock - t >= readyAfter).length;
const show = what => console.log(
  `t=${String(clock).padStart(2)} old=${old} new=${newPods.length}` +
  ` ready=${ready()} total=${old + newPods.length}` +
  ` avail=${old + ready()} ${what}`);

console.log(`maxTotal=${maxTotal} minAvailable=${minAvail}`);
show("start");
while (old > 0 || ready() < R) {
  // Rule 1: create new Pods while total < maxTotal.
  const up = Math.min(maxTotal - old - newPods.length, R - newPods.length);
  if (up > 0) {
    for (let i = 0; i < up; i++) newPods.push(clock);
    show(`+${up} new`);
  }
  // Rule 2: remove old Pods only if available stays >= minAvail.
  const notReady = newPods.length - ready();
  const down = Math.min(old, old + newPods.length - minAvail - notReady);
  if (down > 0) { old -= down; show(`-${down} old`); }
  if (up <= 0 && down <= 0) clock++;     // nothing allowed: wait
}
show("done");

$ kubectl rollout undo deploy/web
deployment.apps/web rolled back

$ kubectl annotate deploy/web kubernetes.io/change-cause="back to 1.27" --overwrite
deployment.apps/web annotated

$ kubectl rollout status deploy/web
…
deployment "web" successfully rolled out

$ kubectl get rs
NAME             DESIRED   CURRENT   READY   AGE
web-84c4bfd5dd   0         0         0       18s
web-df7bd6557    4         4         4       42s

$ kubectl rollout history deploy/web
deployment.apps/web
REVISION  CHANGE-CAUSE
2         image 1.28
3         back to 1.27

$ kubectl rollout pause deploy/web
deployment.apps/web paused

$ kubectl set image deploy/web nginx=nginx:1.28-alpine
deployment.apps/web image updated

$ kubectl set resources deploy/web -c=nginx --limits=memory=80Mi
deployment.apps/web resource requirements updated

$ kubectl get rs
NAME             DESIRED   CURRENT   READY   AGE
web-6f799dff6    0         0         0       15s
web-6f8459f7c8   0         0         0       75s
web-84c4bfd5dd   0         0         0       105s
web-df7bd6557    4         4         4       2m9s

$ kubectl rollout status deploy/web --timeout=4s
Waiting for deployment "web" rollout to finish: 0 out of 4 new replicas have been updated...
error: timed out waiting for the condition

$ kubectl rollout undo deploy/web
error: you cannot rollback a paused deployment; resume it first with 'kubectl rollout resume' and try again

$ kubectl rollout resume deploy/web
deployment.apps/web resumed

$ kubectl get rs
NAME             DESIRED   CURRENT   READY   AGE
web-6f799dff6    0         0         0       26s
web-6f8459f7c8   0         0         0       86s
web-74f8dbbc47   4         4         4       7s
web-84c4bfd5dd   0         0         0       116s
web-df7bd6557    0         0         0       2m20s

$ kubectl patch deploy web -p '{"spec":{"revisionHistoryLimit":1}}'
deployment.apps/web patched

$ kubectl get rs
NAME             DESIRED   CURRENT   READY   AGE
web-59d9464878   0         0         0       75s
web-74f8dbbc47   4         4         4       109s

$ kubectl rollout undo deploy/web --to-revision=1
error: unable to find specified revision 1 in history

$ kubectl set image deploy/web nginx=nginx:1.27-no-such-tag
deployment.apps/web image updated

$ kubectl get rs                    # ৯ সেকেন্ড পরে
NAME             DESIRED   CURRENT   READY   AGE
web-6f8459f7c8   2         2         0       9s
web-84c4bfd5dd   0         0         0       39s
web-df7bd6557    3         3         3       63s

$ kubectl get pods -l app=web
NAME                   READY   STATUS         RESTARTS   AGE
web-6f8459f7c8-w984k   0/1     ErrImagePull   0          9s
web-6f8459f7c8-wzvgs   0/1     ErrImagePull   0          9s
web-df7bd6557-fx6mv    1/1     Running        0          29s
web-df7bd6557-gwn44    1/1     Running        0          29s
web-df7bd6557-hcknr    1/1     Running        0          26s

$ kubectl get deploy web
NAME   READY   UP-TO-DATE   AVAILABLE   AGE
web    3/4     2            3           63s

# Service-এ ২০টি অনুরোধ পাঠানো হলো, ফল গুনে:
     20   HTTP/1.1 200 OK  Server: nginx/1.27.5

$ kubectl rollout status deploy/web
Waiting for deployment "web" rollout to finish: 2 out of 4 new replicas have been updated...
error: deployment "web" exceeded its progress deadline

# kubectl get deploy web -o jsonpath দিয়ে condition:
Available=True MinimumReplicasAvailable: Deployment has minimum availability.
Progressing=False ProgressDeadlineExceeded: ReplicaSet "web-6f8459f7c8" has timed out progressing.

# kubectl describe pod web-6f8459f7c8-w984k (ছাঁটা)
Warning  Failed   …  kubelet  Failed to pull image "nginx:1.27-no-such-tag": … not found
Warning  Failed   …  kubelet  Error: ErrImagePull
Normal   BackOff  …  kubelet  Back-off pulling image "nginx:1.27-no-such-tag"
Warning  Failed   …  kubelet  Error: ImagePullBackOff

// A rollout whose new Pods may never become Ready.
// progressDeadlineSeconds counts time since the LAST progress.
function run(name, R, s, u, deadline, readyAfter) {
  let old = R, pods = [], clock = 0, last = 0, ready = 0, stuck = false;
  while (clock <= 200 && !stuck && (old > 0 || ready < R)) {
    const nowReady = pods.filter(t => clock - t >= readyAfter).length;
    const up = Math.min(R + s - old - pods.length, R - pods.length);
    const fresh = nowReady > ready;
    for (let i = 0; i < up; i++) pods.push(clock);
    ready = nowReady;
    const down = Math.min(old, old + pods.length - (R - u) - (pods.length - ready));
    if (down > 0) old -= down;
    if (up > 0 || down > 0 || fresh) last = clock;
    if (clock - last >= deadline) stuck = true;
    else if (up <= 0 && down <= 0) clock++;
  }
  const avail = old + ready;
  console.log(`${name}: t=${clock}s old=${old} new=${pods.length} ` +
    `ready=${ready} avail=${avail}/${R} ` +
    (stuck ? "ProgressDeadlineExceeded" : "complete"));
}
const never = Infinity;
run("bad image, 1+1   ", 4, 1, 1, 40, never);
run("bad image, 1+0   ", 4, 1, 0, 40, never);
run("bad image, 0+1   ", 4, 0, 1, 40, never);
run("bad image, 3+2   ", 10, 3, 2, 40, never);
run("slow Pods (30s)  ", 4, 1, 0, 40, 30);
run("slower Pods (45s)", 4, 1, 0, 40, 45);

$ kubectl patch deploy web --type=json -p='[{"op":"replace",
  "path":"/spec/template/spec/containers/0/readinessProbe/httpGet/path",
  "value":"/no-such-page"}]'
deployment.apps/web patched

$ kubectl get pods -l app=web
NAME                  READY   STATUS    RESTARTS   AGE
web-6f799dff6-d8b87   0/1     Running   0          10s
web-6f799dff6-wnf6d   0/1     Running   0          10s
web-df7bd6557-fx6mv   1/1     Running   0          90s
web-df7bd6557-gwn44   1/1     Running   0          90s
web-df7bd6557-hcknr   1/1     Running   0          87s

$ kubectl get deploy web
NAME   READY   UP-TO-DATE   AVAILABLE   AGE
web    3/4     2            3           2m4s

# kubectl describe pod থেকে (ছাঁটা):
Warning  Unhealthy  …  kubelet  Readiness probe failed: HTTP probe failed with statuscode: 404

# Service-এ ২০টি অনুরোধ:
     20   HTTP/1.1 200 OK  Server: nginx/1.27.5

$ kubectl patch deploy web -p '{"spec":{"minReadySeconds":10,"progressDeadlineSeconds":120}}'
deployment.apps/web patched

$ kubectl set image deploy/web nginx=nginx:1.27-alpine
deployment.apps/web image updated

$ kubectl get deploy web          # ৫ সেকেন্ড পরে
NAME   READY   UP-TO-DATE   AVAILABLE   AGE
web    5/4     2            3           2m53s

$ kubectl get rs
NAME             DESIRED   CURRENT   READY   AGE
web-59d9464878   2         2         2       6s
…
web-74f8dbbc47   3         3         3       40s
…

$ kubectl get deploy web          # আরও ৫ সেকেন্ড পরে
NAME   READY   UP-TO-DATE   AVAILABLE   AGE
web    5/4     2            3           2m57s

// Rollout time: the quick formula against a small simulation.
function simulate(R, s, u, ready, minReady) {
  const wait = ready + minReady;
  let old = R, pods = [], clock = 0;
  const av = () => pods.filter(t => clock - t >= wait).length;
  while (old > 0 || av() < R) {
    const up = Math.min(R + s - old - pods.length, R - pods.length);
    for (let i = 0; i < up; i++) pods.push(clock);
    const down = Math.min(old, old + pods.length - (R - u) - (pods.length - av()));
    if (down > 0) old -= down;
    if (up <= 0 && down <= 0) clock++;
  }
  return clock;
}
const estimate = (R, s, u, ready, minReady) =>
  Math.ceil(R / (s + u)) * (ready + minReady);

console.log("R  surge unav ready minReady | formula sim");
for (const [R, s, u, ready, minReady] of [
  [4, 1, 1, 3, 0], [4, 1, 1, 3, 10], [4, 1, 0, 3, 0], [10, 3, 2, 20, 0],
  [10, 1, 0, 20, 10], [100, 25, 25, 30, 0], [100, 1, 0, 30, 0],
]) {
  console.log(`${R} ${s} ${u} ${ready} ${minReady} | ` +
    `${estimate(R, s, u, ready, minReady)}s ` +
    `${simulate(R, s, u, ready, minReady)}s`);
}

$ kubectl patch deploy web -p '{"spec":{"strategy":{"type":"Recreate","rollingUpdate":null}}}'
deployment.apps/web patched

$ kubectl set image deploy/web nginx=nginx:1.28-alpine
deployment.apps/web image updated

$ kubectl get deploy web          # ১ সেকেন্ড পরে
NAME   READY   UP-TO-DATE   AVAILABLE   AGE
web    0/4     0            0           3m19s

$ kubectl get pods -l app=web
NAME                   READY   STATUS      RESTARTS   AGE
web-59d9464878-n2d77   0/1     Completed   0          32s
web-59d9464878-pfhr4   0/1     Completed   0          19s
web-59d9464878-sf7b2   0/1     Completed   0          32s
web-59d9464878-svt7m   0/1     Completed   0          19s

# rollout-এর সময় চলা ২০০টি অনুরোধের ফল:
recreate ok=188 bad=12

apiVersion: apps/v1
kind: Deployment
metadata:
  name: api
spec:
  replicas: 4
  selector:
    matchLabels:
      app: api
  strategy:
    type: RollingUpdate
    rollingUpdate:
      maxSurge: 1
      maxUnavailable: 0
  template:
    metadata:
      labels:
        app: api
    spec:
      terminationGracePeriodSeconds: 30
      containers:
        - name: nginx
          image: nginx:1.27-alpine
          command: ["sh", "-c", "sleep 4; exec nginx -g 'daemon off;'"]
          ports:
            - containerPort: 80
          readinessProbe:
            httpGet:
              path: /
              port: 80
            periodSeconds: 1
            failureThreshold: 1
          lifecycle:
            preStop:
              exec:
                command: ["sleep", "5"]
          resources:
            requests:
              cpu: 20m
              memory: 16Mi
            limits:
              cpu: 200m
              memory: 64Mi

apiVersion: policy/v1
kind: PodDisruptionBudget
metadata:
  name: web-pdb
spec:
  minAvailable: 3
  selector:
    matchLabels:
      app: web

$ kubectl get pdb
NAME      MIN AVAILABLE   MAX UNAVAILABLE   ALLOWED DISRUPTIONS   AGE
web-pdb   3               N/A               1                     3s

# প্রথম Pod সরানোর অনুরোধ (Eviction API, কমান্ড ছাঁটা):
{"kind":"Status","apiVersion":"v1","metadata":{},"status":"Success","code":201}

# সাথে সাথে দ্বিতীয় Pod:
Error from server (TooManyRequests): Cannot evict pod as it would violate the pod's disruption budget.

$ kubectl get pdb
NAME      MIN AVAILABLE   MAX UNAVAILABLE   ALLOWED DISRUPTIONS   AGE
web-pdb   3               N/A               0                     3s

$ kubectl get pods -l app=web
NAME                   READY   STATUS              RESTARTS   AGE
web-59d9464878-24jrj   1/1     Terminating         0          7s
web-59d9464878-45qbz   1/1     Running             0          7s
web-59d9464878-d6sb9   1/1     Running             0          10s
web-59d9464878-qk4kn   1/1     Running             0          10s
web-59d9464878-r67t5   0/1     ContainerCreating   0          0s

$ kubectl set image deploy/web nginx=nginx:1.28-alpine     # বদলি Ready হওয়ার পরে
deployment.apps/web image updated

$ kubectl get pdb
NAME      MIN AVAILABLE   MAX UNAVAILABLE   ALLOWED DISRUPTIONS   AGE
web-pdb   3               N/A               0                     11s

apiVersion: apps/v1
kind: DaemonSet
metadata:
  name: log-agent
spec:
  selector:
    matchLabels:
      app: log-agent
  template:
    metadata:
      labels:
        app: log-agent
    spec:
      containers:
        - name: agent
          image: alpine:3.20
          command: ["sh", "-c", "sleep 3000"]

$ kubectl get ds
NAME        DESIRED   CURRENT   READY   UP-TO-DATE   AVAILABLE   NODE SELECTOR   AGE
log-agent   1         1         1       1            1           <none>          6s

apiVersion: batch/v1
kind: Job
metadata:
  name: always-fails
spec:
  backoffLimit: 2
  template:
    spec:
      restartPolicy: Never
      containers:
        - name: task
          image: alpine:3.20
          command: ["sh", "-c", "echo processing; exit 1"]
---
apiVersion: batch/v1
kind: CronJob
metadata:
  name: nightly-report
spec:
  schedule: "0 2 * * *"
  timeZone: "Asia/Dhaka"
  concurrencyPolicy: Forbid
  successfulJobsHistoryLimit: 2
  failedJobsHistoryLimit: 1
  jobTemplate:
    spec:
      backoffLimit: 1
      template:
        spec:
          restartPolicy: Never
          containers:
            - name: report
              image: alpine:3.20
              command: ["sh", "-c", "echo report done"]

$ kubectl get jobs
NAME                      STATUS     COMPLETIONS   DURATION   AGE
always-fails              Failed     0/1           51s        51s
nightly-report-29889810   Complete   1/1           5s         34s
third-try                 Complete   1/1           7s         51s

$ kubectl get pods -l job-name=always-fails
NAME                 READY   STATUS   RESTARTS   AGE
always-fails-jn25s   0/1     Error    0          51s
always-fails-lghzs   0/1     Error    0          18s
always-fails-q8dzx   0/1     Error    0          39s

$ kubectl get pods -l job-name=third-try
NAME              READY   STATUS      RESTARTS   AGE
third-try-zpczm   0/1     Completed   2          51s

$ kubectl logs -l job-name=third-try
try 3