Skip to content
Branch: master
Find file History
Fetching latest commit…
Cannot retrieve the latest commit at this time.
Permalink
Type Name Latest commit message Commit time
..
Failed to load latest commit information.
101_simple-rs.yaml
201_simple-rs-5.yaml
202_simple-rs-50.yaml
203_simple-rs-pods.yaml
301_simple-blue-pods.yaml
302_simple-blue-rs.yaml
303_simple-red-pods.yaml
304_simple-rs-nocolor-update.yaml
305_simple-nocolor-rs.yaml
306_simple-orange-pods.yaml
307_simple-orange-rs.yaml
400_probes-rs-readiness.yaml
401_probes-rs-readiness-ko.yaml
402_probes-rs-liveness.yaml
403_probes-rs-liveness-ko.yaml
501_probes-images-rs.yaml
502_probes-images-rs-update-image-ko.yaml
503_probes-images-rs-6-update-image-ko.yaml
504_probes-images-rs-9-update-image-ok.yaml
505_probes-images-rs-6-update-image-ok.yaml
506_probes-images-rs-update-selector.yaml
507_probes-images-v2.0-rs.yaml Complete the ReplicaSet README May 11, 2019
508_probes-images-scale-down-both-rs.yaml
509_probes-images-scale-up-both-rs.yaml
README.md

README.md

ReplicaSets

Introduction

A ReplicaSet is defined with fields, including a selector that specifies how to identify Pods it can acquire, a number of replicas indicating how many Pods it should be maintaining, and a pod template specifying the data of new Pods it should create to meet the number of replicas criteria. A ReplicaSet then fulfills its purpose by creating and deleting Pods as needed to reach the desired number. When a ReplicaSet needs to create new Pods, it uses its Pod template.

Learn more

Some notes

  • Kubernetes is fast, specially when working with lightweight containers, so to see what is happening, we'll need to run kubectl commands fast.

That's why we will run the kubectl command to apply changes and if works, immediately after, a kubectl get with the -w, --watch flag:

-w, --watch=false: After listing/getting the requested object, watch for changes. Uninitialized objects are excluded if no object name is provided.

This will allow us to get the information on what is happening just after the command is applied:

$ kubectl run --generator 'run-pod/v1' --image bash:5.0 --restart Never sleepy -
- sleep 10 && kubectl get pods sleepy -w
pod/sleepy created
NAME     READY   STATUS              RESTARTS   AGE
sleepy   0/1     ContainerCreating   0          0s
sleepy   1/1     Running             0          1s
sleepy   0/1     Completed           0          12s

This command creates a Pod called sleepy with a container using the image bash:5.0 and running a sleep 10. Once the first command completes, the kubectl get pods sleepy -w, will watch a Pod called sleepy and print a new line for each change in the Pod status.

1 - Create a ReplicaSet

Create the ReplicaSet object:

apiVersion: apps/v1
kind: ReplicaSet
metadata:
  name: simple
spec:
  replicas: 1
  selector:
    matchLabels:
      app: simple
  template:
    metadata:
      labels:
        app: simple
    spec:
      containers:
      
      - name: app
        image: raelga/cats:gatet

Apply the yaml:

$ kubectl apply -f 101_simple-rs.yaml
replicaset.apps/simple created
$ kubectl get pods
NAME           READY   STATUS              RESTARTS   AGE
simple-hfsgg   0/1     ContainerCreating   0          3s
$ kubectl get pods
NAME           READY   STATUS    RESTARTS   AGE
simple-hfsgg   1/1     Running   0          32s
$ kubectl get rs simple
NAME     DESIRED   CURRENT   READY   AGE
simple   1         1         1       29s

2 - Scaling ReplicaSets

Double the numbers of replicas with kubectl scale

$ kubectl scale rs/simple --replicas 2
replicaset.extensions/simple scaled
$ k get rs simple
NAME     DESIRED   CURRENT   READY   AGE
simple   2         2         2       104s
$ kubectl get pods
NAME           READY   STATUS    RESTARTS   AGE
simple-hfsgg   1/1     Running   0          74s
simple-kj8k8   1/1     Running   0          3s

Scale back to 1 replica

$ kubectl scale rs/simple --replicas 1
replicaset.extensions/simple scaled
$ kubectl get pods
NAME           READY   STATUS        RESTARTS   AGE
simple-s8n2f   1/1     Terminating   0          48s
simple-hfsgg   1/1     Running       0          119s
$ kubectl get pods
NAME           READY   STATUS    RESTARTS   AGE
simple-hfsgg   1/1     Running   0          125s

Update the ReplicaSet with the yaml definition

We also add information about the resources needed by the app container of the pod, to let the scheduler know how many pods can fit into a node.

$ kubectl diff -f 201_simple-rs-5.yaml 
diff -u -N /tmp/LIVE-860683615/apps.v1.ReplicaSet.default.simple /tmp/MERGED-116908338/apps.v1.ReplicaSet.default.simple
--
- /tmp/LIVE-860683615/apps.v1.ReplicaSet.default.simple       2019-05-11 11:05:20.751504492 +0000
+++ /tmp/MERGED-116908338/apps.v1.ReplicaSet.default.simple     2019-05-11 11:05:20.767506423 +0000
@@ -5,14 +5,14 @@
     kubectl.kubernetes.io/last-applied-configuration: |
       {"apiVersion":"apps/v1","kind":"ReplicaSet","metadata":{"annotations":{},"name":"simple","namespace":"default"},"spec":{"replicas":1,"selector":{"matchLabels":{"app":"simple"}},"template":{"metadata":{"labels":{"app":"simple"}},"spec":{"containers":[{"image":"raelga/cats:gatet","name":"app"}]}}}}
   creationTimestamp: "2019-05-11T11:01:50Z"

-  generation: 3
+  generation: 4
   name: simple
   namespace: default
   resourceVersion: "1560"
   selfLink: /apis/apps/v1/namespaces/default/replicasets/simple
   uid: 2db859ed-73dc-11e9-9a36-eefc5b75fd0d
 spec:

-  replicas: 1
+  replicas: 5
   selector:
     matchLabels:
       app: simple
@@ -26,7 +26,13 @@
       
       - image: raelga/cats:gatet
         imagePullPolicy: IfNotPresent
         name: app

-        resources: {}
+        resources:
+          limits:
+            cpu: "1"
+            memory: 100Mi
+          requests:
+            cpu: 100m
+            memory: 50Mi
         terminationMessagePath: /dev/termination-log
         terminationMessagePolicy: File
       dnsPolicy: ClusterFirst
exit status 1
$ kubectl apply -f 201_simple-rs-5.yaml && kubectl get pods -w
replicaset.apps/simple configured
NAME           READY   STATUS              RESTARTS   AGE
simple-dgksp   0/1     ContainerCreating   0          0s
simple-hfsgg   1/1     Running             0          4m58s
simple-hm6gj   0/1     ContainerCreating   0          0s
simple-htdzs   0/1     ContainerCreating   0          0s
simple-rnbt6   0/1     ContainerCreating   0          0s
simple-htdzs   1/1     Running             0          2s
simple-hm6gj   1/1     Running             0          3s
simple-dgksp   1/1     Running             0          4s
simple-rnbt6   1/1     Running             0          4s
$ kubectl get pods
NAME           READY   STATUS    RESTARTS   AGE
simple-dgksp   1/1     Running   0          10s
simple-hfsgg   1/1     Running   0          5m8s
simple-hm6gj   1/1     Running   0          10s
simple-htdzs   1/1     Running   0          10s
simple-rnbt6   1/1     Running   0          10s

Scale to 50 replicas

$ kubectl diff -f 202_simple-rs-50.yaml
diff -u -N /tmp/LIVE-082216881/apps.v1.ReplicaSet.default.simple /tmp/MERGED-524798300/apps.v1.ReplicaSet.default.simple
--
- /tmp/LIVE-082216881/apps.v1.ReplicaSet.default.simple       2019-05-11 11:09:06.426174215 +0000
+++ /tmp/MERGED-524798300/apps.v1.ReplicaSet.default.simple     2019-05-11 11:09:06.440175724 +0000
@@ -5,14 +5,14 @@
     kubectl.kubernetes.io/last-applied-configuration: |
       {"apiVersion":"apps/v1","kind":"ReplicaSet","metadata":{"annotations":{},"name":"simple","namespace":"default"},"spec":{"replicas":5,"selector":{"matchLabels":{"app":"simple"}},"template":{"metadata":{"labels":{"app":"simple"}},"spec":{"containers":[{"image":"raelga/cats:gatet","name":"app","resources":{"limits":{"cpu":"1","memory":"100Mi"},"requests":{"cpu":"50m","memory":"50Mi"}}}]}}}}
   creationTimestamp: "2019-05-11T11:01:50Z"

-  generation: 9
+  generation: 10
   name: simple
   namespace: default
   resourceVersion: "2292"
   selfLink: /apis/apps/v1/namespaces/default/replicasets/simple
   uid: 2db859ed-73dc-11e9-9a36-eefc5b75fd0d
 spec:

-  replicas: 5
+  replicas: 50
   selector:
     matchLabels:
       app: simple
exit status 1
$ k apply -f 202_simple-rs-50.yaml && kubectl get rs simple -w
replicaset.apps/simple configured
NAME     DESIRED   CURRENT   READY   AGE
simple   50        5         5       12m
simple   50        5         5       12m
simple   50        50        5       12m
simple   50        50        6       12m
simple   50        50        7       12m
simple   50        50        8       12m
simple   50        50        9       12m
simple   50        50        10      12m
simple   50        50        11      12m
simple   50        50        12      12m
simple   50        50        13      12m
simple   50        50        14      12m
simple   50        50        15      12m
simple   50        50        16      12m
simple   50        50        17      12m
simple   50        50        18      12m
simple   50        50        19      13m
simple   50        50        20      13m
simple   50        50        21      13m
simple   50        50        22      13m
simple   50        50        23      13m
simple   50        50        24      13m
simple   50        50        25      13m
simple   50        50        26      13m
simple   50        50        27      13m
simple   50        50        28      13m
simple   50        50        29      13m
simple   50        50        30      13m
simple   50        50        31      13m
simple   50        50        32      13m
simple   50        50        33      13m
simple   50        50        34      13m
simple   50        50        35      13m

Check the pods that are not Running:

$ kubectl get pods --field-selector='status.phase!=Running'
NAME           READY   STATUS    RESTARTS   AGE
simple-7fbc4   0/1     Pending   0          39s
simple-9qsdq   0/1     Pending   0          39s
simple-b2xlw   0/1     Pending   0          39s
simple-bfdrl   0/1     Pending   0          39s
simple-dsnb7   0/1     Pending   0          39s
simple-f2jd6   0/1     Pending   0          38s
simple-gn48j   0/1     Pending   0          39s
simple-htlsv   0/1     Pending   0          39s
simple-qdhhh   0/1     Pending   0          39s
simple-rjvpm   0/1     Pending   0          39s
simple-rnbs8   0/1     Pending   0          39s
simple-s9msf   0/1     Pending   0          39s
simple-snkvj   0/1     Pending   0          38s
simple-th97b   0/1     Pending   0          39s
simple-tzz9n   0/1     Pending   0          39s
$ kubectl describe $(kubectl get pods --field-selector='status.phase!=Running' --output name | head -n1)
Name:               simple-7fbc4
Namespace:          default
Priority:           0
PriorityClassName:  <none>
Node:               <none>
Labels:             app=simple
Annotations:        <none>
Status:             Pending
IP:                 
Controlled By:      ReplicaSet/simple
Containers:
  app:
    Image:      raelga/cats:gatet
    Port:       <none>
    Host Port:  <none>
    Limits:
      cpu:     1
      memory:  100Mi
    Requests:
      cpu:        100m
      memory:     50Mi
    Environment:  <none>
    Mounts:
      /var/run/secrets/kubernetes.io/serviceaccount from default-token-5mrjx (ro)
Conditions:
  Type           Status
  PodScheduled   False 
Volumes:
  default-token-5mrjx:
    Type:        Secret (a volume populated by a Secret)
    SecretName:  default-token-5mrjx
    Optional:    false
QoS Class:       Burstable
Node-Selectors:  <none>
Tolerations:     node.kubernetes.io/not-ready:NoExecute for 300s
                 node.kubernetes.io/unreachable:NoExecute for 300s
Events:
  Type     Reason            Age                From               Message
  ---
  -     -----
  -            ---
  -               ---
  -               -------
  Warning  FailedScheduling  59s (x2 over 59s)  default-scheduler  0/2 nodes are available: 2 Insufficient cpu.

We can see that the Pod cannot be scheduled due to insufficient cpus in the node pool.

Warning FailedScheduling 41s (x6 over 3m58s) default-scheduler 0/2 nodes are available: 2 Insufficient cpu.

Scale down back to 5 replicas

$ kubectl scale rs/simple --replicas 5 && kubectl get rs/simple -w
replicaset.extensions/simple scaled
NAME     DESIRED   CURRENT   READY   AGE
simple   5         50        35      34m
simple   5         50        35      34m
simple   5         5         5       34m
$ kubectl get pods  
NAME           READY   STATUS    RESTARTS   AGE
simple-hfsgg   1/1     Running   0          34m
simple-hzhpc   1/1     Running   0          22m
simple-wjdf9   1/1     Running   0          22m
simple-xgq2x   1/1     Running   0          22m
simple-xh5hm   1/1     Running   0          22m

3 - Selectors and Pods

Deploy some blue pods

---
apiVersion: v1
kind: Pod
metadata:
  name: simple-blue-pod-1
  labels:
    app: simple
    color: blue
spec:
  containers:
    
    - name: app
      image: raelga/cats:liam
...
$ kubectl apply -f 301_simple-blue-pods.yaml && kubectl get pods -w
pod/simple-blue-pod-1 created
pod/simple-blue-pod-2 created
pod/simple-blue-pod-3 created
NAME                READY   STATUS        RESTARTS   AGE
simple-blue-pod-1   0/1     Terminating   0          1s
simple-blue-pod-2   0/1     Terminating   0          0s
simple-blue-pod-3   0/1     Terminating   0          0s
simple-hfsgg        1/1     Running       0          107m
simple-hzhpc        1/1     Running       0          95m
simple-wjdf9        1/1     Running       0          95m
simple-xgq2x        1/1     Running       0          95m
simple-xh5hm        1/1     Running       0          95m
simple-blue-pod-3   0/1     Terminating   0          2s
simple-blue-pod-3   0/1     Terminating   0          2s
simple-blue-pod-1   0/1     Terminating   0          5s
simple-blue-pod-1   0/1     Terminating   0          5s
simple-blue-pod-2   0/1     Terminating   0          11s
simple-blue-pod-2   0/1     Terminating   0          11s
$ k get pods
NAME           READY   STATUS    RESTARTS   AGE
simple-hfsgg   1/1     Running   0          108m
simple-hzhpc   1/1     Running   0          96m
simple-wjdf9   1/1     Running   0          96m
simple-xgq2x   1/1     Running   0          96m
simple-xh5hm   1/1     Running   0          96m

The new blue pods get Terminated! Why??

$ k describe rs/simple
Name:         simple
Namespace:    default
Selector:     app=simple
Labels:       <none>
Annotations:  kubectl.kubernetes.io/last-applied-configuration:
                {"apiVersion":"apps/v1","kind":"ReplicaSet","metadata":{"annotations":{},"name":"simple","namespace":"default"},"spec":{"replicas":50,"sel...
Replicas:     5 current / 5 desired
Pods Status:  5 Running / 0 Waiting / 0 Succeeded / 0 Failed
Pod Template:
  Labels:  app=simple
  Containers:
   app:
    Image:      raelga/cats:gatet
    Port:       <none>
    Host Port:  <none>
    Limits:
      cpu:     1
      memory:  100Mi
    Requests:
      cpu:        100m
      memory:     50Mi
    Environment:  <none>
    Mounts:       <none>
  Volumes:        <none>
Events:
  Type    Reason            Age   From                   Message
  ---
  -    -----
  -            ---
  -  ---
  -                   -------
  Normal  SuccessfulDelete  71s   replicaset-controller  Deleted pod: simple-blue-pod-1
  Normal  SuccessfulDelete  71s   replicaset-controller  Deleted pod: simple-blue-pod-2
  Normal  SuccessfulDelete  71s   replicaset-controller  Deleted pod: simple-blue-pod-3

We can see that the ReplicaSet terminated those Pods:

  Normal  SuccessfulDelete  71s   replicaset-controller  Deleted pod: simple-blue-pod-1
  Normal  SuccessfulDelete  71s   replicaset-controller  Deleted pod: simple-blue-pod-2
  Normal  SuccessfulDelete  71s   replicaset-controller  Deleted pod: simple-blue-pod-3

Deploy a blue ReplicaSet

$ kubectl apply -f 302_simple-blue-rs.yaml && kubectl get pods -w
replicaset.apps/simple-blue created
NAME                READY   STATUS              RESTARTS   AGE
simple-blue-cff2b   0/1     Pending             0          0s
simple-blue-dp26t   0/1     Pending             0          0s
simple-blue-kfl7z   0/1     Pending             0          0s
simple-blue-kr46r   0/1     Pending             0          0s
simple-blue-nng6p   0/1     ContainerCreating   0          0s
simple-hfsgg        1/1     Running             0          112m
simple-hzhpc        1/1     Running             0          100m
simple-wjdf9        1/1     Running             0          100m
simple-xgq2x        1/1     Running             0          100m
simple-xh5hm        1/1     Running             0          100m
simple-blue-kfl7z   0/1     ContainerCreating   0          0s
simple-blue-dp26t   0/1     ContainerCreating   0          0s
simple-blue-cff2b   0/1     ContainerCreating   0          0s
simple-blue-kr46r   0/1     ContainerCreating   0          0s
simple-blue-cff2b   1/1     Running             0          2s
simple-blue-kfl7z   1/1     Running             0          3s
simple-blue-nng6p   1/1     Running             0          4s
simple-blue-kr46r   1/1     Running             0          4s
simple-blue-dp26t   1/1     Running             0          4s
$ kubectl get pods
NAME                READY   STATUS    RESTARTS   AGE
simple-blue-cff2b   1/1     Running   0          11s
simple-blue-dp26t   1/1     Running   0          11s
simple-blue-kfl7z   1/1     Running   0          11s
simple-blue-kr46r   1/1     Running   0          11s
simple-blue-nng6p   1/1     Running   0          11s
simple-hfsgg        1/1     Running   0          112m
simple-hzhpc        1/1     Running   0          100m
simple-wjdf9        1/1     Running   0          100m
simple-xgq2x        1/1     Running   0          100m
simple-xh5hm        1/1     Running   0          100m
$ kubectl get rs
NAME          DESIRED   CURRENT   READY   AGE
simple        5         5         5       112m
simple-blue   5         5         5       34s

Now the Pods stay, but why?

The selector is matching sets of pods, from most restrictive to less restrictive. So rs/simple with manage all pods with label: app: simple that doesn't match any other replicaset.

Run a red pod

---
apiVersion: v1
kind: Pod
metadata:
  name: simple-red-pod-1
  labels:
    app: simple
    color: red
spec:
  containers:
    
    - name: app
      image: raelga/cats:liam

This new Pods, have a color label. Will be deleted?

$ kubectl apply -f 303_simple-red-pods.yaml && kubectl get pods -w
pod/simple-red-pod-1 created
NAME                READY   STATUS        RESTARTS   AGE
simple-blue-cff2b   1/1     Running       0          75s
simple-blue-dp26t   1/1     Running       0          75s
simple-blue-kfl7z   1/1     Running       0          75s
simple-blue-kr46r   1/1     Running       0          75s
simple-blue-nng6p   1/1     Running       0          75s
simple-hfsgg        1/1     Running       0          113m
simple-hzhpc        1/1     Running       0          101m
simple-red-pod-1    0/1     Terminating   0          0s
simple-wjdf9        1/1     Running       0          101m
simple-xgq2x        1/1     Running       0          101m
simple-xh5hm        1/1     Running       0          101m
simple-red-pod-1    0/1     Terminating   0          7s
simple-red-pod-1    0/1     Terminating   0          7s
$ kubectl describe rs simple | grep simple-red-pod
  Normal  SuccessfulDelete  46s    replicaset-controller  Deleted pod: simple-red-pod-1

The red Pod is killed because:

  • Has the app: simple label selector

  • There is no any ReplicaSet for the color: red pods

This pods matches simple ReplicaSet selector and there is already 5 pods pods matching the selector, so should be terminated.

ReplicaSet for non-colored pods only

The new selector for the rs/simple-nocolor will be a combination of a label and an expresion:

  selector:
    matchLabels:
      app: simple
    matchExpressions:
      
      - { key: color, operator: DoesNotExist }

Let's try to update the simple ReplicaSet:

$ kubectl diff -f 304_simple-rs-nocolor-update.yaml 
The ReplicaSet "simple" is invalid: spec.selector: Invalid value: v1.LabelSelector{MatchLabels:map[string]string{"app":"simple"}, MatchExpressions:[]v1.LabelSelectorRequirement{v1.LabelSelectorRequirement{Key:"color", Operator:"DoesNotExist", Values:[]string(nil)}}}: field is immutable

``ReplicaSetsare defined by the.spec.selectors` and are inmmutable.`

Let's create a new ReplicaSet then:

$ kubectl apply -f 305_simple-nocolor-rs.yaml
replicaset.apps/simple-nocolor created
$ kubectl get rs
NAME             DESIRED   CURRENT   READY   AGE
simple           5         5         5       121m
simple-blue      5         5         5       9m38s
simple-nocolor   5         5         5       16s

Let's create some orange Pods:

$ kubectl apply -f 306_simple-orange-pods.yaml && kubectl get pods -w -l color=orange
pod/simple-orange-pod-1 created
pod/simple-orange-pod-2 created
pod/simple-orange-pod-3 created
NAME                  READY   STATUS        RESTARTS   AGE
simple-orange-pod-1   0/1     Terminating   0          0s
simple-orange-pod-2   0/1     Terminating   0          0s
simple-orange-pod-3   0/1     Terminating   0          0s

They are still getting delted by the rs/simple controller.

$ kubectl describe rs simple | grep orange        
  Normal  SuccessfulDelete  33s (x3 over 89s)  replicaset-controller  Deleted pod: simple-orange-pod-1
  Normal  SuccessfulDelete  33s (x3 over 89s)  replicaset-controller  Deleted pod: simple-orange-pod-2
  Normal  SuccessfulDelete  33s (x3 over 89s)  replicaset-controller  Deleted pod: simple-orange-pod-3

Let's remove the simple ReplicaSet then:

$ kubectl delete rs/simple            
replicaset.extensions "simple" deleted
$ kubectl get rs          
NAME             DESIRED   CURRENT   READY   AGE
simple-blue      5         5         5       12m
simple-nocolor   5         5         5       3m32s

Let's create again the orange Pods.

$ kubectl apply -f 306_simple-orange-pods.yaml && kubectl get pods -w -l color=orange
pod/simple-orange-pod-1 created
pod/simple-orange-pod-2 created
pod/simple-orange-pod-3 created
NAME                  READY   STATUS              RESTARTS   AGE
simple-orange-pod-1   0/1     ContainerCreating   0          1s
simple-orange-pod-2   0/1     ContainerCreating   0          1s
simple-orange-pod-3   0/1     ContainerCreating   0          0s
simple-orange-pod-2   1/1     Running             0          3s
simple-orange-pod-3   1/1     Running             0          4s
simple-orange-pod-1   1/1     Running             0          7s
$ kubectl get pods
NAME                   READY   STATUS    RESTARTS   AGE
simple-blue-cff2b      1/1     Running   0          13m
simple-blue-dp26t      1/1     Running   0          13m
simple-blue-kfl7z      1/1     Running   0          13m
simple-blue-kr46r      1/1     Running   0          13m
simple-blue-nng6p      1/1     Running   0          13m
simple-nocolor-pcf6d   1/1     Running   0          3m58s
simple-nocolor-qfn9l   1/1     Running   0          3m58s
simple-nocolor-rlkdl   1/1     Running   0          3m58s
simple-nocolor-txglk   1/1     Running   0          3m58s
simple-nocolor-vmxm5   1/1     Running   0          3m58s
simple-orange-pod-1    1/1     Running   0          11s
simple-orange-pod-2    1/1     Running   0          11s
simple-orange-pod-3    1/1     Running   0          10s

Let's acquire those fancy orange pods

kubectl apply -f 307_simple-orange-rs.yaml
replicaset.apps/simple-orange created
$ kubectl get pods -l color=orange
NAME                  READY   STATUS    RESTARTS   AGE
simple-orange-k72nq   1/1     Running   0          23s
simple-orange-mlh9h   1/1     Running   0          23s
simple-orange-pod-1   1/1     Running   0          2m14s
simple-orange-pod-2   1/1     Running   0          2m14s
simple-orange-pod-3   1/1     Running   0          2m13s
simple-orange-vkclj   1/1     Running   0          23s

As you can see, the ReplicaSet only created the required pods to have 6 replicas:

$ kubectl describe rs simple-orange | grep SuccessfulCreate
  Normal  SuccessfulCreate  67s   replicaset-controller  Created pod: simple-orange-mlh9h
  Normal  SuccessfulCreate  67s   replicaset-controller  Created pod: simple-orange-k72nq
  Normal  SuccessfulCreate  67s   replicaset-controller  Created pod: simple-orange-vkclj

Remove a pod from the orange replicaset

$ kubectl patch pod simple-orange-pod-1 --type='json' --patch='[{"op":"replace", "path":"/metadata/labels/color", "value":"pink"}]' && kubectl get pods -w -l color=orange
pod/simple-orange-pod-1 patched
NAME                  READY   STATUS              RESTARTS   AGE
simple-orange-5s295   1/1     Running             0          4m2s
simple-orange-6nd67   0/1     ContainerCreating   0          0s
simple-orange-gzwjs   1/1     Running             0          4m3s
simple-orange-pod-2   1/1     Running             0          4m6s
simple-orange-pod-3   1/1     Running             0          4m6s
simple-orange-v9lsm   1/1     Running             0          4m2s
simple-orange-6nd67   1/1     Running             0          2s

The simple-orange-pod-1 is no longer part of the ReplicaSet and the simple-orange-6nd67 has been created to ensure that there are 6 replicas running.

$ kubectl patch pod simple-orange-pod-2 --type='json' --patch='[{"op":"replace", "path":"/metadata/labels/color", "value":"pink"}]' && kubectl get rs simple-orange -w
pod/simple-orange-pod-2 patched
NAME            DESIRED   CURRENT   READY   AGE
simple-orange   6         6         5       8m13s
simple-orange   6         6         6       8m15s
$ kubectl get pods -l color=pink
NAME                  READY   STATUS    RESTARTS   AGE
simple-orange-pod-1   1/1     Running   0          9m21s
simple-orange-pod-2   1/1     Running   0          9m21s

Clean up

$ kubectl delete rs simple-blue simple-nocolor simple-orange
replicaset.extensions "simple-blue" deleted
replicaset.extensions "simple-nocolor" deleted
replicaset.extensions "simple-orange" deleted
$ kubectl get pods
NAME                  READY   STATUS    RESTARTS   AGE
simple-orange-pod-1   1/1     Running   0          9m50s
simple-orange-pod-2   1/1     Running   0          9m50s
$ kubectl delete pods -l color=pink 
pod "simple-orange-pod-1" deleted
pod "simple-orange-pod-2" deleted
$ kubectl get pods
No resources found.

4 - Container probes

Readiness probe

Let's add a ReadinessProbe, on the port 80:

  • Wait 10 secs before start probing

  • Each 5 seconds, check the probe

  • Mark as healthy after 3 consecutive OK checks

...
    spec:
      containers:
      
      - name: app
        image: raelga/cats:neu
        readinessProbe:
          tcpSocket:
            port: 80
          initialDelaySeconds: 10
          periodSeconds: 5
          successThreshold: 3
$ kubectl apply -f 400_probes-rs-readiness.yaml && kubectl get pods -w -l app=probes
replicaset.apps/probes created
NAME           READY   STATUS              RESTARTS   AGE
probes-m2l5h   0/1     ContainerCreating   0          0s
probes-x56j5   0/1     ContainerCreating   0          0s
probes-m2l5h   0/1     Running             0          2s
probes-x56j5   0/1     Running             0          4s
probes-x56j5   1/1     Running             0          24s
probes-m2l5h   1/1     Running             0          24s

After about 25 seconds, the Pods became READY:

  • Scheduling time (From Pending to Running)

  • 10 seconds of initialDelaySeconds

  • 15 seconds (5 periodSeconds x 3 successThreshold)

Deploy some Pods with failing ReadinessProbes:

$ kubectl apply -f 401_probes-rs-readiness-ko.yaml && kubectl get pods -l app=probes -w
replicaset.apps/probes configured
NAME           READY   STATUS              RESTARTS   AGE
probes-csqfn   0/1     ContainerCreating   0          0s
probes-m2l5h   1/1     Running             0          51s
probes-np7vg   0/1     Pending             0          0s
probes-x56j5   1/1     Running             0          51s
probes-np7vg   0/1     ContainerCreating   0          0s
probes-csqfn   0/1     Running             0          2s
probes-np7vg   0/1     Running             0          3s

They are not getting READY at all:

$ kubectl get pods -l app=probes
NAME           READY   STATUS    RESTARTS   AGE
probes-csqfn   0/1     Running   0          86s
probes-m2l5h   1/1     Running   0          2m17s
probes-np7vg   0/1     Running   0          86s
probes-x56j5   1/1     Running   0          2m17s

And if we look the status of the READY 0/1 pods, we'll see the reason in the Events: section:

$ kubectl describe pods $(kubectl get pods | sed -n 's:\(\S\+\)\s\+0/1.*:\1:p') | grep Warning
  Warning  Unhealthy  2s (x17 over 82s)  kubelet, cnbcn-k8s-study-jam-np-fw7y  Readiness probe failed: dial tcp 10.244.1.225:81: connect: connection refused
  Warning  Unhealthy  0s (x17 over 80s)  kubelet, cnbcn-k8s-study-jam-np-fw7x  Readiness probe failed: dial tcp 10.244.0.55:81: connect: connection refused
$ kubectl describe pods $(kubectl get pods | sed -n 's:\(\S\+\)\s\+0/1.*:\1:p')
...
Events:
  Type     Reason     Age                From                                  Message
  ---
  -     -----
  -     ---
  -               ---
  -                                  -------
  Normal   Scheduled  109s               default-scheduler                     Successfully assigned default/probes-np7vg to cnbcn-k8s-study-jam-np-fw7x
  Normal   Pulled     107s               kubelet, cnbcn-k8s-study-jam-np-fw7x  Container image "raelga/cats:neu" already present on machine
  Normal   Created    107s               kubelet, cnbcn-k8s-study-jam-np-fw7x  Created container
  Normal   Started    107s               kubelet, cnbcn-k8s-study-jam-np-fw7x  Started container
  Warning  Unhealthy  3s (x19 over 93s)  kubelet, cnbcn-k8s-study-jam-np-fw7x  Readiness probe failed: dial tcp 10.244.0.55:81: connect: connection refused

The good thing, is the failing Pods never get READY and won't receive any traffic.

Liveness probe

Let's add a LivenessProbe, on the port 80:

  • Wait 10 secs before start probing

  • Each 5 seconds, check the probe

  • Mark as unhealthy after 2 consecutive failed checks

...
    spec:
      containers:
      
      - name: app
        image: raelga/cats:neu
        livenessProbe:
          httpGet:
            path: /
            port: 80
          initialDelaySeconds: 10
          periodSeconds: 5
          failureThreshold: 2
$ kubectl apply -f 402_probes-rs-liveness.yaml && kubectl get pods -l app=probes -w
replicaset.apps/probes configured
NAME           READY   STATUS              RESTARTS   AGE
probes-csqfn   0/1     Running             0          2m19s
probes-hhn5x   0/1     ContainerCreating   0          0s
probes-m2l5h   1/1     Running             0          3m10s
probes-np7vg   0/1     Running             0          2m19s
probes-ntl94   0/1     ContainerCreating   0          0s
probes-x56j5   1/1     Running             0          3m10s
probes-hhn5x   1/1     Running             0          2s
probes-ntl94   1/1     Running             0          2s

The Pods using the LivenessProbe get READY in a few seconds, just when the container starts and responds the first probe request. Why don't we check the failureThreshold instead of the successThreshold?

Because this only checks that the application is live, and if fails, will restart the container instead of keep re-checking untill the application to become ready again.

Let's launch a failing LivenessProbe to see this behavior:

$ kubectl apply -f 403_probes-rs-liveness-ko.yaml && kubectl get pods -l app=probes -w
replicaset.apps/probes configured
NAME           READY   STATUS              RESTARTS   AGE
probes-csqfn   0/1     Running             0          7m10s
probes-fzbgb   0/1     Pending             0          0s
probes-hhn5x   1/1     Running             0          4m51s
probes-m2l5h   1/1     Running             0          8m1s
probes-np7vg   0/1     Running             0          7m10s
probes-ntl94   1/1     Running             0          4m51s
probes-tmlln   0/1     ContainerCreating   0          0s
probes-x56j5   1/1     Running             0          8m1s
probes-fzbgb   0/1     ContainerCreating   0          0s
probes-fzbgb   0/1     Running             0          3s
probes-tmlln   0/1     Running             0          3s
probes-tmlln   0/1     Running             1          19s
probes-fzbgb   0/1     Running             1          20s
probes-tmlln   1/1     Running             1          33s
probes-tmlln   0/1     Running             2          34s
probes-fzbgb   1/1     Running             1          34s
probes-fzbgb   0/1     Running             2          40s
probes-tmlln   1/1     Running             2          48s
probes-tmlln   0/1     Running             3          50s
probes-fzbgb   1/1     Running             2          54s
probes-fzbgb   0/1     Running             3          61s
probes-tmlln   1/1     Running             3          63s
probes-tmlln   0/1     CrashLoopBackOff    3          65s
probes-fzbgb   1/1     Running             3          74s
probes-fzbgb   0/1     Running             4          80s
probes-tmlln   0/1     Running             4          94s
probes-fzbgb   1/1     Running             4          94s
probes-fzbgb   0/1     CrashLoopBackOff    4          100s
probes-tmlln   1/1     Running             4          108s
probes-tmlln   0/1     Running             5          109s
probes-tmlln   1/1     Running             5          2m3s
probes-tmlln   0/1     CrashLoopBackOff    5          2m5s
probes-fzbgb   0/1     Running             5          2m28s
probes-fzbgb   1/1     Running             5          2m39s
probes-fzbgb   0/1     CrashLoopBackOff    5          2m45s
probes-tmlln   0/1     Running             6          3m27s
probes-tmlln   1/1     Running             6          3m38s
probes-tmlln   0/1     CrashLoopBackOff    6          3m44s
probes-fzbgb   0/1     Running             6          4m14s
probes-fzbgb   1/1     Running             6          4m24s
probes-fzbgb   0/1     CrashLoopBackOff    6          4m31s

A CrashloopBackOff means that we have a pod starting, crashing, starting again, and then crashing again. Failed containers that are restarted by the kubelet are restarted with an exponential back-off delay (10s, 20s, 40s …) capped at five minutes, and is reset after ten minutes of successful execution.

Clean up

$ kubectl delete rs probes && kubectl get pods -w -l app=probes
replicaset.extensions "probes" deleted
NAME           READY   STATUS        RESTARTS   AGE
probes-csqfn   0/1     Terminating   0          29m
probes-fzbgb   0/1     Terminating   11         22m
probes-hhn5x   1/1     Terminating   0          27m
probes-m2l5h   1/1     Terminating   0          30m
probes-np7vg   0/1     Terminating   0          29m
probes-ntl94   1/1     Terminating   0          27m
probes-tmlln   0/1     Terminating   11         22m
probes-x56j5   1/1     Terminating   0          30m
probes-x56j5   0/1     Terminating   0          30m
probes-ntl94   0/1     Terminating   0          27m
probes-np7vg   0/1     Terminating   0          30m
probes-csqfn   0/1     Terminating   0          30m
probes-hhn5x   0/1     Terminating   0          27m
probes-m2l5h   0/1     Terminating   0          30m
probes-m2l5h   0/1     Terminating   0          30m
probes-tmlln   0/1     Terminating   11         22m
probes-tmlln   0/1     Terminating   11         22m
probes-csqfn   0/1     Terminating   0          30m
probes-csqfn   0/1     Terminating   0          30m
probes-hhn5x   0/1     Terminating   0          27m
probes-hhn5x   0/1     Terminating   0          27m
probes-fzbgb   0/1     Terminating   11         22m
probes-fzbgb   0/1     Terminating   11         22m
probes-m2l5h   0/1     Terminating   0          30m
probes-m2l5h   0/1     Terminating   0          30m
probes-x56j5   0/1     Terminating   0          30m
probes-x56j5   0/1     Terminating   0          30m
probes-ntl94   0/1     Terminating   0          27m
probes-ntl94   0/1     Terminating   0          27m
probes-np7vg   0/1     Terminating   0          30m
probes-np7vg   0/1     Terminating   0          30m
$ kubectl get pods
No resources found.

5 - Manual rolling update

Deploy the initial ReplicaSet

$ kubectl apply -f 501_probes-images-rs.yaml && kubectl get pods -l app=probes-images -w
replicaset.apps/probes-images created
NAME                  READY   STATUS    RESTARTS   AGE
probes-images-dtfv4   0/1     Pending   0          0s
probes-images-jnck4   0/1     Pending   0          0s
probes-images-dtfv4   0/1     Pending   0          0s
probes-images-h5wsz   0/1     Pending   0          0s
probes-images-h5wsz   0/1     Pending   0          0s
probes-images-jnck4   0/1     Pending   0          0s
probes-images-dtfv4   0/1     ContainerCreating   0          0s
probes-images-jnck4   0/1     ContainerCreating   0          0s
probes-images-h5wsz   0/1     ContainerCreating   0          0s
probes-images-h5wsz   0/1     Running             0          2s
probes-images-jnck4   0/1     Running             0          3s
probes-images-dtfv4   0/1     Running             0          3s
probes-images-h5wsz   1/1     Running             0          7s
probes-images-jnck4   1/1     Running             0          7s
probes-images-dtfv4   1/1     Running             0          10s

Let's list the pods running, including the name of the first container image, using custom-columns output format:

-o custom-columns=NAME:.metadata.name,IMAGE:.spec.containers[0].image,STATUS:.status.phase,STARTED:.status.startTime

$ kubectl get pods -l app=probes-images -o custom-columns=NAME:.metadata.name,IMAGE:.spec.containers[0].image,STATUS:.status.phase,STARTED:.status.startTime
NAME                  IMAGE             STATUS    STARTED
probes-images-dtfv4   raelga/cats:neu   Running   2019-05-11T16:20:36Z
probes-images-h5wsz   raelga/cats:neu   Running   2019-05-11T16:20:36Z
probes-images-jnck4   raelga/cats:neu   Running   2019-05-11T16:20:36Z

Update the ReplicaSet pod template

Let's update the ReplicaSet Pod template and include a new template, that will fail the ReadinessProbe.

$ kubectl diff -f 502_probes-images-rs-update-image-ko.yaml 
diff -u -N /tmp/LIVE-839052409/apps.v1.ReplicaSet.default.probes-images /tmp/MERGED-353118084/apps.v1.ReplicaSet.default.probes-images
--
- /tmp/LIVE-839052409/apps.v1.ReplicaSet.default.probes-images        2019-05-11 16:21:15.916702182 +0000
+++ /tmp/MERGED-353118084/apps.v1.ReplicaSet.default.probes-images      2019-05-11 16:21:15.929703344 +0000
@@ -5,7 +5,7 @@
     kubectl.kubernetes.io/last-applied-configuration: |
       {"apiVersion":"apps/v1","kind":"ReplicaSet","metadata":{"annotations":{},"labels":{"app":"probes-images"},"name":"probes-images","namespace":"default"},"spec":{"replicas":3,"selector":{"matchLabels":{"app":"probes-images"}},"template":{"metadata":{"labels":{"app":"probes-images"}},"spec":{"containers":[{"image":"raelga/cats:neu","livenessProbe":{"failureThreshold":2,"httpGet":{"path":"/","port":80},"initialDelaySeconds":10,"periodSeconds":5},"name":"app","readinessProbe":{"httpGet":{"path":"/","port":80},"initialDelaySeconds":2}}]}}}}
   creationTimestamp: "2019-05-11T16:20:36Z"

-  generation: 1
+  generation: 2
   labels:
     app: probes-images
   name: probes-images
@@ -25,7 +25,7 @@
         app: probes-images
     spec:
       containers:

-      
- image: raelga/cats:neu
+      
- image: raelga/cats:blanca
         imagePullPolicy: IfNotPresent
         livenessProbe:
           failureThreshold: 2
@@ -41,7 +41,7 @@
         readinessProbe:
           failureThreshold: 3
           httpGet:

-            path: /
+            path: /bad-endpoint
             port: 80
             scheme: HTTP
           initialDelaySeconds: 2
exit status 1
$ kubectl apply -f 502_probes-images-rs-update-image-ko.yaml && kubectl get pods -l app=probes-images -w
replicaset.apps/probes-images configured
NAME                  READY   STATUS    RESTARTS   AGE
probes-images-dtfv4   1/1     Running   0          56s
probes-images-h5wsz   1/1     Running   0          56s
probes-images-jnck4   1/1     Running   0          56s

Nothing happens, as the ReplicaSet only cares for the number of replicas, the new Pod template will be used for creating new Pods when required, but has no effect on the existing ReplicaSet Pods.

$ kubectl get rs probes-images
NAME            DESIRED   CURRENT   READY   AGE
probes-images   3         3         3       20m

Let's increase the number of replicas to 6:

$ kubectl diff -f 503_probes-images-rs-6-update-image-ko.yaml 
diff -u -N /tmp/LIVE-648056706/apps.v1.ReplicaSet.default.probes-images /tmp/MERGED-796264697/apps.v1.ReplicaSet.default.probes-images
--
- /tmp/LIVE-648056706/apps.v1.ReplicaSet.default.probes-images        2019-05-11 16:23:07.606688192 +0000
+++ /tmp/MERGED-796264697/apps.v1.ReplicaSet.default.probes-images      2019-05-11 16:23:07.619689354 +0000
@@ -5,7 +5,7 @@
     kubectl.kubernetes.io/last-applied-configuration: |
       {"apiVersion":"apps/v1","kind":"ReplicaSet","metadata":{"annotations":{},"labels":{"app":"probes-images"},"name":"probes-images","namespace":"default"},"spec":{"replicas":3,"selector":{"matchLabels":{"app":"probes-images"}},"template":{"metadata":{"labels":{"app":"probes-images"}},"spec":{"containers":[{"image":"raelga/cats:blanca","livenessProbe":{"failureThreshold":2,"httpGet":{"path":"/","port":80},"initialDelaySeconds":10,"periodSeconds":5},"name":"app","readinessProbe":{"httpGet":{"path":"/bad-endpoint","port":80},"initialDelaySeconds":2}}]}}}}
   creationTimestamp: "2019-05-11T16:20:36Z"

-  generation: 2
+  generation: 3
   labels:
     app: probes-images
   name: probes-images
@@ -14,7 +14,7 @@
   selfLink: /apis/apps/v1/namespaces/default/replicasets/probes-images
   uid: b57546b3-7408-11e9-9a36-eefc5b75fd0d
 spec:

-  replicas: 3
+  replicas: 6
   selector:
     matchLabels:
       app: probes-images
exit status 1

Now the start but never get READY, as they are not passing the ReadinessProbe:

$ kubectl apply -f 503_probes-images-rs-6-update-image-ko.yaml && kubectl get pods -l app=probes-images -w
replicaset.apps/probes-images configured
NAME                  READY   STATUS              RESTARTS   AGE
probes-images-42vbb   0/1     ContainerCreating   0          0s
probes-images-dhs88   0/1     ContainerCreating   0          0s
probes-images-dtfv4   1/1     Running             0          3m21s
probes-images-fhrlp   0/1     ContainerCreating   0          0s
probes-images-h5wsz   1/1     Running             0          3m21s
probes-images-jnck4   1/1     Running             0          3m21s
probes-images-dhs88   0/1     Running             0          2s
probes-images-42vbb   0/1     Running             0          2s
probes-images-fhrlp   0/1     Running             0          2s

At this point, the old image is still running and serving requests. The new version is failing, but never gets traffic, as is not passing the ReadinessProbe.

$ kubectl get rs probes-images
NAME            DESIRED   CURRENT   READY   AGE
probes-images   6         6         3       4m25s
$ kubectl get pods -l app=probes-images -o custom-columns=NAME:.metadata.name,IMAGE:.spec.containers[0].image,STATUS:.status.phase,STARTED:.status.startTime
NAME                  IMAGE                STATUS    STARTED
probes-images-42vbb   raelga/cats:blanca   Running   2019-05-11T16:23:57Z
probes-images-dhs88   raelga/cats:blanca   Running   2019-05-11T16:23:57Z
probes-images-dtfv4   raelga/cats:neu      Running   2019-05-11T16:20:36Z
probes-images-fhrlp   raelga/cats:blanca   Running   2019-05-11T16:23:57Z
probes-images-h5wsz   raelga/cats:neu      Running   2019-05-11T16:20:36Z
probes-images-jnck4   raelga/cats:neu      Running   2019-05-11T16:20:36Z

We can see the ReadinessProbe error using kubectl describe:

$ kubectl describe pods -l app=probes-images | grep Warning
  Warning  Unhealthy  9s (x19 over 3m9s)  kubelet, cnbcn-k8s-study-jam-np-fw7x  Readiness probe failed: HTTP probe failed with statuscode: 404
  Warning  Unhealthy  3s (x19 over 3m3s)  kubelet, cnbcn-k8s-study-jam-np-fw7y  Readiness probe failed: HTTP probe failed with statuscode: 404
  Warning  Unhealthy  1s (x20 over 3m11s)  kubelet, cnbcn-k8s-study-jam-np-fw7x  Readiness probe failed: HTTP probe failed with statuscode: 404

Update the ReplicaSet Pod template with the fixed ReadinessProbe

$ kubectl diff -f 504_probes-images-rs-9-update-image-ok.yaml 
diff -u -N /tmp/LIVE-210527442/apps.v1.ReplicaSet.default.probes-images /tmp/MERGED-362151945/apps.v1.ReplicaSet.default.probes-images
--
- /tmp/LIVE-210527442/apps.v1.ReplicaSet.default.probes-images        2019-05-11 16:30:04.733986406 +0000
+++ /tmp/MERGED-362151945/apps.v1.ReplicaSet.default.probes-images      2019-05-11 16:30:04.746987569 +0000
@@ -5,7 +5,7 @@
     kubectl.kubernetes.io/last-applied-configuration: |
       {"apiVersion":"apps/v1","kind":"ReplicaSet","metadata":{"annotations":{},"labels":{"app":"probes-images"},"name":"probes-images","namespace":"default"},"spec":{"replicas":6,"selector":{"matchLabels":{"app":"probes-images"}},"template":{"metadata":{"labels":{"app":"probes-images"}},"spec":{"containers":[{"image":"raelga/cats:blanca","livenessProbe":{"failureThreshold":2,"httpGet":{"path":"/","port":80},"initialDelaySeconds":10,"periodSeconds":5},"name":"app","readinessProbe":{"httpGet":{"path":"/bad-endpoint","port":80},"initialDelaySeconds":2}}]}}}}
   creationTimestamp: "2019-05-11T16:20:36Z"

-  generation: 3
+  generation: 4
   labels:
     app: probes-images
   name: probes-images
@@ -14,7 +14,7 @@
   selfLink: /apis/apps/v1/namespaces/default/replicasets/probes-images
   uid: b57546b3-7408-11e9-9a36-eefc5b75fd0d
 spec:

-  replicas: 6
+  replicas: 9
   selector:
     matchLabels:
       app: probes-images
@@ -41,7 +41,7 @@
         readinessProbe:
           failureThreshold: 3
           httpGet:

-            path: /bad-endpoint
+            path: /
             port: 80
             scheme: HTTP
           initialDelaySeconds: 2
exit status 1
$ kubectl apply -f 504_probes-images-rs-9-update-image-ok.yaml && kubectl get rs probes-images -w
replicaset.apps/probes-images configured
NAME            DESIRED   CURRENT   READY   AGE
probes-images   9         9         3       11m
probes-images   9         9         4       11m
probes-images   9         9         5       11m
probes-images   9         9         6       11m

The ReplicaSet now has 9 replicas, 6 of them with the new image:

$ kubectl get pods -l app=probes-images -o custom-columns=NAME:.metadata.name,IMAGE:.spec.containers[0].image,STATUS:.status.phase,STARTED:.status.startTime --sort-by=.status.startTime
NAME                  IMAGE                STATUS    STARTED
probes-images-jnck4   raelga/cats:neu      Running   2019-05-11T16:20:36Z
probes-images-dtfv4   raelga/cats:neu      Running   2019-05-11T16:20:36Z
probes-images-h5wsz   raelga/cats:neu      Running   2019-05-11T16:20:36Z
probes-images-dhs88   raelga/cats:blanca   Running   2019-05-11T16:23:57Z
probes-images-fhrlp   raelga/cats:blanca   Running   2019-05-11T16:23:57Z
probes-images-42vbb   raelga/cats:blanca   Running   2019-05-11T16:23:57Z
probes-images-7g9cc   raelga/cats:blanca   Running   2019-05-11T16:32:04Z
probes-images-r87pj   raelga/cats:blanca   Running   2019-05-11T16:32:04Z
probes-images-ssxzx   raelga/cats:blanca   Running   2019-05-11T16:32:04Z

And the last 3 replicas, with the fixed ReadinessProbe, are READY:

$ kubectl get pods --sort-by=.status.startTime
NAME                  READY   STATUS    RESTARTS   AGE
probes-images-jnck4   1/1     Running   0          12m
probes-images-dtfv4   1/1     Running   0          12m
probes-images-h5wsz   1/1     Running   0          12m
probes-images-dhs88   0/1     Running   0          9m7s
probes-images-fhrlp   0/1     Running   0          9m7s
probes-images-42vbb   0/1     Running   0          9m7s
probes-images-7g9cc   1/1     Running   0          60s
probes-images-r87pj   1/1     Running   0          60s
probes-images-ssxzx   1/1     Running   0          60s

The new ReplicaSet template is creating healthy Pods!

Clean up the failing versions and the old ones

Now, let's clean up the failing Pods by scaling back to 6!

$ kubectl scale rs/probes-images --replicas 6
replicaset.extensions/probes-images scaled
$ kubectl get pods
NAME                  READY   STATUS        RESTARTS   AGE
probes-images-42vbb   0/1     Terminating   0          13m
probes-images-7g9cc   1/1     Running       0          5m27s
probes-images-dhs88   0/1     Terminating   0          13m
probes-images-dtfv4   1/1     Running       0          16m
probes-images-fhrlp   0/1     Terminating   0          13m
probes-images-h5wsz   1/1     Running       0          16m
probes-images-jnck4   1/1     Running       0          16m
probes-images-r87pj   1/1     Running       0          5m27s
probes-images-ssxzx   1/1     Running       0          5m27s
$ kubectl get pods -l app=probes-images -o custom-columns=NAME:.metadata.name,IMAGE:.spec.containers[0].image,STATUS:.status.phase,STARTED:.status.startTime --sort-by=.status.startTime
NAME                  IMAGE                STATUS    STARTED
probes-images-dtfv4   raelga/cats:neu      Running   2019-05-11T16:20:36Z
probes-images-h5wsz   raelga/cats:neu      Running   2019-05-11T16:20:36Z
probes-images-jnck4   raelga/cats:neu      Running   2019-05-11T16:20:36Z
probes-images-7g9cc   raelga/cats:blanca   Running   2019-05-11T16:32:04Z
probes-images-r87pj   raelga/cats:blanca   Running   2019-05-11T16:32:04Z
probes-images-ssxzx   raelga/cats:blanca   Running   2019-05-11T16:32:04Z

Let's now clean the old version by scaling back to 3!

$ kubectl scale rs/probes-images --replicas 3    
replicaset.extensions/probes-images scaled
$ kubectl get pods
NAME                  READY   STATUS        RESTARTS   AGE
probes-images-dtfv4   1/1     Running       0          20m
probes-images-h5wsz   1/1     Running       0          20m
probes-images-jnck4   1/1     Running       0          20m
probes-images-r87pj   0/1     Terminating   0          9m13s
probes-images-ssxzx   0/1     Terminating   0          9m13s

Wow! It removed the new ones instead of the old ones! Why?

In the first scenario, the ReplicaSet saw that some of the replicas weren't READY and removed then. But now, both the ones with the new image and the ones with the old are healthy, so in that scenario, will remove always the newest ones.

pkg/controller/replicaset/replica_set.go:getPodsToDelete

func getPodsToDelete(filteredPods []*v1.Pod, diff int) []*v1.Pod {
	// No need to sort pods if we are about to delete all of them.
	// diff will always be <= len(filteredPods), so not need to handle > case.
	if diff < len(filteredPods) {
		// Sort the pods in the order such that not-ready < ready, unscheduled
		// < scheduled, and pending < running. This ensures that we delete pods
		// in the earlier stages whenever possible.
		sort.Sort(controller.ActivePods(filteredPods))
	}
	return filteredPods[:diff]
}

And the code for ActivePods:

pkg/controller/controller_utils.go:ActivePods

func (s ActivePods) Less(i, j int) bool {
	// 1. Unassigned < assigned
	// If only one of the pods is unassigned, the unassigned one is smaller
	if s[i].Spec.NodeName != s[j].Spec.NodeName && (len(s[i].Spec.NodeName) == 0 || len(s[j].Spec.NodeName) == 0) {
		return len(s[i].Spec.NodeName) == 0
	}
	// 2. PodPending < PodUnknown < PodRunning
	m := map[v1.PodPhase]int{v1.PodPending: 0, v1.PodUnknown: 1, v1.PodRunning: 2}
	if m[s[i].Status.Phase] != m[s[j].Status.Phase] {
		return m[s[i].Status.Phase] < m[s[j].Status.Phase]
	}
	// 3. Not ready < ready
	// If only one of the pods is not ready, the not ready one is smaller
	if podutil.IsPodReady(s[i]) != podutil.IsPodReady(s[j]) {
		return !podutil.IsPodReady(s[i])
	}
	// TODO: take availability into account when we push minReadySeconds information from deployment into pods,
	//       see https://github.com/kubernetes/kubernetes/issues/22065
	// 4. Been ready for empty time < less time < more time
	// If both pods are ready, the latest ready one is smaller
	if podutil.IsPodReady(s[i]) && podutil.IsPodReady(s[j]) && !podReadyTime(s[i]).Equal(podReadyTime(s[j])) {
		return afterOrZero(podReadyTime(s[i]), podReadyTime(s[j]))
	}
	// 5. Pods with containers with higher restart counts < lower restart counts
	if maxContainerRestarts(s[i]) != maxContainerRestarts(s[j]) {
		return maxContainerRestarts(s[i]) > maxContainerRestarts(s[j])
	}
	// 6. Empty creation time pods < newer pods < older pods
	if !s[i].CreationTimestamp.Equal(&s[j].CreationTimestamp) {
		return afterOrZero(&s[i].CreationTimestamp, &s[j].CreationTimestamp)
	}
	return false
}

Let's scale back to 6, so we will have the 3 replicas with the old version and 3 replicas with the new template:

diff -u -N /tmp/LIVE-571681951/apps.v1.ReplicaSet.default.probes-images /tmp/MERGED-052727154/apps.v1.ReplicaSet.default.probes-images
--
- /tmp/LIVE-571681951/apps.v1.ReplicaSet.default.probes-images        2019-05-11 16:58:39.800320854 +0000
+++ /tmp/MERGED-052727154/apps.v1.ReplicaSet.default.probes-images      2019-05-11 16:58:39.823322911 +0000
@@ -5,7 +5,7 @@
     kubectl.kubernetes.io/last-applied-configuration: |
       {"apiVersion":"apps/v1","kind":"ReplicaSet","metadata":{"annotations":{},"labels":{"app":"probes-images"},"name":"probes-images","namespace":"default"},"spec":{"replicas":6,"selector":{"matchLabels":{"app":"probes-images"}},"template":{"metadata":{"labels":{"app":"probes-images"}},"spec":{"containers":[{"image":"raelga/cats:blanca","livenessProbe":{"failureThreshold":2,"httpGet":{"path":"/","port":80},"initialDelaySeconds":10,"periodSeconds":5},"name":"app","readinessProbe":{"httpGet":{"path":"/","port":80},"initialDelaySeconds":2}}]}}}}
   creationTimestamp: "2019-05-11T16:20:36Z"

-  generation: 20
+  generation: 21
   labels:
     app: probes-images
   name: probes-images
@@ -14,7 +14,7 @@
   selfLink: /apis/apps/v1/namespaces/default/replicasets/probes-images
   uid: b57546b3-7408-11e9-9a36-eefc5b75fd0d
 spec:

-  replicas: 3
+  replicas: 6
   selector:
     matchLabels:
       app: probes-images
@@ -23,6 +23,7 @@
       creationTimestamp: null
       labels:
         app: probes-images
+        version: v2.0
     spec:
       containers:
       
       - image: raelga/cats:blanca
exit status 1
$ kubectl apply -f 505_probes-images-rs-6-update-image-ok.yaml && kubectl get pods -l app=probes-images -o custom-columns=NAME:.metadata.name,IMAGE:.spec.containers[0].image,STATUS:.status.phase,STARTED:.status.startTime --sort-by=.status.startTime -w
replicaset.apps/probes-images configured
NAME                  IMAGE                STATUS    STARTED
probes-images-dtfv4   raelga/cats:neu      Running   2019-05-11T16:20:36Z
probes-images-h5wsz   raelga/cats:neu      Running   2019-05-11T16:20:36Z
probes-images-jnck4   raelga/cats:neu      Running   2019-05-11T16:20:36Z
probes-images-pbltv   raelga/cats:blanca   Pending   2019-05-11T17:00:12Z
probes-images-sv698   raelga/cats:blanca   Pending   2019-05-11T17:00:12Z
probes-images-vbrqc   raelga/cats:blanca   Pending   2019-05-11T17:00:12Z
probes-images-pbltv   raelga/cats:blanca   Running   2019-05-11T17:00:12Z
probes-images-sv698   raelga/cats:blanca   Running   2019-05-11T17:00:12Z
probes-images-vbrqc   raelga/cats:blanca   Running   2019-05-11T17:00:12Z
probes-images-sv698   raelga/cats:blanca   Running   2019-05-11T17:00:12Z
probes-images-pbltv   raelga/cats:blanca   Running   2019-05-11T17:00:12Z
probes-images-vbrqc   raelga/cats:blanca   Running   2019-05-11T17:00:12Z
$ kubectl get pods -l app=probes-images -o custom-columns=NAME:.metadata.name,IMAGE:.spec.containers[0].image,STATUS:.status.phase,STARTED:.status.startTime --sort-by=.status.startTime
NAME                  IMAGE                STATUS    STARTED
probes-images-dtfv4   raelga/cats:neu      Running   2019-05-11T16:20:36Z
probes-images-h5wsz   raelga/cats:neu      Running   2019-05-11T16:20:36Z
probes-images-jnck4   raelga/cats:neu      Running   2019-05-11T16:20:36Z
probes-images-pbltv   raelga/cats:blanca   Running   2019-05-11T17:00:12Z
probes-images-sv698   raelga/cats:blanca   Running   2019-05-11T17:00:12Z
probes-images-vbrqc   raelga/cats:blanca   Running   2019-05-11T17:00:12Z

As we added a new label to the template, is easy to identify the new replicas:

$ kubectl get pods -l app=probes-images,version=v2.0  
NAME                  READY   STATUS    RESTARTS   AGE
probes-images-pbltv   1/1     Running   0          55s
probes-images-sv698   1/1     Running   0          55s
probes-images-vbrqc   1/1     Running   0          55s

And now, we can remove the old version and the ReplicaSet will replace the terminated pods with new ones, using the ReplicaSet template:

$ kubectl get pods -l app=probes-images,version!=v2.0
NAME                  READY   STATUS    RESTARTS   AGE
probes-images-dtfv4   1/1     Running   0          41m
probes-images-h5wsz   1/1     Running   0          41m
probes-images-jnck4   1/1     Running   0          41m
$ kubectl delete  pods -l app=probes-images,version!=v2.0
pod "probes-images-dtfv4" deleted
pod "probes-images-h5wsz" deleted
pod "probes-images-jnck4" deleted
$  kubectl get pods -l app=probes-images -o custom-columns=NAME:.metadata.name,IMAGE:.spec.containers[0].image,STATUS:.status.phase,STARTED:.status.startTime --sort-by=.status.startTime   
NAME                  IMAGE                STATUS    STARTED
probes-images-pbltv   raelga/cats:blanca   Running   2019-05-11T17:00:12Z
probes-images-sv698   raelga/cats:blanca   Running   2019-05-11T17:00:12Z
probes-images-vbrqc   raelga/cats:blanca   Running   2019-05-11T17:00:12Z
probes-images-62v48   raelga/cats:blanca   Running   2019-05-11T17:02:53Z
probes-images-bs6z9   raelga/cats:blanca   Running   2019-05-11T17:02:53Z
probes-images-ndhd7   raelga/cats:blanca   Running   2019-05-11T17:02:53Z
$ kubectl get pods -l app=probes-images,version!=v2.0
No resources found.

That worked! The problem, is that now the ReplicaSet selector is not the same as the labels defined in the Pod template, so another if another ReplicaSet is created with the exact match, this ReplicaSet will lose the Pods.

So let's update the ReplicaSet with the new selector!

$ diff -U5 505_probes-images-rs-6-update-image-ok.yaml 506_probes-images-rs-update-selector.yaml 
--
- 505_probes-images-rs-6-update-image-ok.yaml 2019-05-11 16:58:45.738851782 +0000
+++ 506_probes-images-rs-update-selector.yaml   2019-05-11 17:09:35.580954662 +0000
@@ -7,10 +7,11 @@
 spec:
   replicas: 6
   selector:
     matchLabels:
       app: probes-images
+      version: v2.0
   template:
     metadata:
       labels:
         app: probes-images
         version: v2.0
$ kubectl apply -f 506_probes-images-rs-update-selector.yaml
The ReplicaSet "probes-images" is invalid: spec.selector: Invalid value: v1.LabelSelector{MatchLabels:map[string]string{"app":"probes-images", "version":"v2.0"}, MatchExpressions:[]v1.LabelSelectorRequirement(nil)}: field is immutable

As we already saw before, the ReplicaSet selector field is immutable, so we need to create a new ReplicaSet.

$ diff -U5 505_probes-images-rs-6-update-image-ok.yaml 507_probes-images-v2.0-rs.yaml 
--
- 505_probes-images-rs-6-update-image-ok.yaml 2019-05-11 16:58:45.738851782 +0000
+++ 507_probes-images-v2.0-rs.yaml      2019-05-11 17:09:20.425599538 +0000
@@ -1,16 +1,17 @@
 apiVersion: apps/v1
 kind: ReplicaSet
 metadata:

-  name: probes-images
+  name: probes-images-v2.0
   labels:
     app: probes-images
 spec:
   replicas: 6
   selector:
     matchLabels:
       app: probes-images
+      version: v2.0
   template:
     metadata:
       labels:
         app: probes-images
         version: v2.0

Let's check the current status:

$ kubectl get rs -l app=probes-images
NAME            DESIRED   CURRENT   READY   AGE
probes-images   6         6         6       52m

Let's add the new ReplicaSet!

$ kubectl apply -f 507_probes-images-v2.0-rs.yaml && kubectl get rs -l app=probes-images -w
replicaset.apps/probes-images-v2.0 unchanged
NAME                 DESIRED   CURRENT   READY   AGE
probes-images        6         6         6       53m
probes-images-v2.0   6         6         2       10s
probes-images-v2.0   6         6         3       10s
probes-images-v2.0   6         6         4       11s
probes-images-v2.0   6         6         5       13s
probes-images-v2.0   6         6         6       14s

We changed the ReplicaSet, so new replicas have been created with the new ReplicaSet name prefix but leave the olds as were generated by another ReplicaSet. The ownership information is stored in the .metadata.ownerReferences array:

$  kubectl get pods -l app=probes-images -o custom-columns=NAME:.metadata.name,LABELS:.metadata.labels,OWNER-TYPE:.metadata.ownerReferences[0].kind,OWNER-NAME:.metadata.ownerReferences[0].name --sort-by=.status.startTime
NAME                       LABELS                                OWNER-TYPE   OWNER-NAME
probes-images-pbltv        map[app:probes-images version:v2.0]   ReplicaSet   probes-images
probes-images-sv698        map[app:probes-images version:v2.0]   ReplicaSet   probes-images
probes-images-vbrqc        map[app:probes-images version:v2.0]   ReplicaSet   probes-images
probes-images-62v48        map[app:probes-images version:v2.0]   ReplicaSet   probes-images
probes-images-bs6z9        map[app:probes-images version:v2.0]   ReplicaSet   probes-images
probes-images-ndhd7        map[app:probes-images version:v2.0]   ReplicaSet   probes-images
probes-images-v2.0-9kpln   map[app:probes-images version:v2.0]   ReplicaSet   probes-images-v2.0
probes-images-v2.0-dnl2k   map[app:probes-images version:v2.0]   ReplicaSet   probes-images-v2.0
probes-images-v2.0-dxj4v   map[app:probes-images version:v2.0]   ReplicaSet   probes-images-v2.0
probes-images-v2.0-hg8xb   map[app:probes-images version:v2.0]   ReplicaSet   probes-images-v2.0
probes-images-v2.0-mmvzr   map[app:probes-images version:v2.0]   ReplicaSet   probes-images-v2.0
probes-images-v2.0-86shb   map[app:probes-images version:v2.0]   ReplicaSet   probes-images-v2.0

We can see, that all the pods have the same labels, but are managed by different ReplicaSets and they will manage differnt set of pods with the same labels:

Let's scale down both ReplicaSets and see what happens:

$ kubectl diff -f 508_probes-images-scale-down-both-rs.yaml 
diff -u -N /tmp/LIVE-427424072/apps.v1.ReplicaSet.default.probes-images /tmp/MERGED-238269447/apps.v1.ReplicaSet.default.probes-images
--
- /tmp/LIVE-427424072/apps.v1.ReplicaSet.default.probes-images        2019-05-11 17:41:33.346631258 +0000
+++ /tmp/MERGED-238269447/apps.v1.ReplicaSet.default.probes-images      2019-05-11 17:41:33.360632510 +0000
@@ -5,7 +5,7 @@
     kubectl.kubernetes.io/last-applied-configuration: |
       {"apiVersion":"apps/v1","kind":"ReplicaSet","metadata":{"annotations":{},"labels":{"app":"probes-images"},"name":"probes-images","namespace":"default"},"spec":{"replicas":6,"selector":{"matchLabels":{"app":"probes-images"}},"template":{"metadata":{"labels":{"app":"probes-images","version":"v2.0"}},"spec":{"containers":[{"image":"raelga/cats:blanca","livenessProbe":{"failureThreshold":2,"httpGet":{"path":"/","port":80},"initialDelaySeconds":10,"periodSeconds":5},"name":"app","readinessProbe":{"httpGet":{"path":"/","port":80},"initialDelaySeconds":2}}]}}}}
   creationTimestamp: "2019-05-11T16:20:36Z"

-  generation: 23
+  generation: 24
   labels:
     app: probes-images
   name: probes-images
@@ -14,7 +14,7 @@
   selfLink: /apis/apps/v1/namespaces/default/replicasets/probes-images
   uid: b57546b3-7408-11e9-9a36-eefc5b75fd0d
 spec:

-  replicas: 6
+  replicas: 2
   selector:
     matchLabels:
       app: probes-images
diff -u -N /tmp/LIVE-427424072/apps.v1.ReplicaSet.default.probes-images-v2.0 /tmp/MERGED-238269447/apps.v1.ReplicaSet.default.probes-images-v2.0
--
- /tmp/LIVE-427424072/apps.v1.ReplicaSet.default.probes-images-v2.0   2019-05-11 17:41:33.486643786 +0000
+++ /tmp/MERGED-238269447/apps.v1.ReplicaSet.default.probes-images-v2.0 2019-05-11 17:41:33.500645039 +0000
@@ -5,7 +5,7 @@
     kubectl.kubernetes.io/last-applied-configuration: |
       {"apiVersion":"apps/v1","kind":"ReplicaSet","metadata":{"annotations":{},"labels":{"app":"probes-images"},"name":"probes-images-v2.0","namespace":"default"},"spec":{"replicas":6,"selector":{"matchLabels":{"app":"probes-images","version":"v2.0"}},"template":{"metadata":{"labels":{"app":"probes-images","version":"v2.0"}},"spec":{"containers":[{"image":"raelga/cats:blanca","livenessProbe":{"failureThreshold":2,"httpGet":{"path":"/","port":80},"initialDelaySeconds":10,"periodSeconds":5},"name":"app","readinessProbe":{"httpGet":{"path":"/","port":80},"initialDelaySeconds":2}}]}}}}
   creationTimestamp: "2019-05-11T17:13:37Z"

-  generation: 1
+  generation: 2
   labels:
     app: probes-images
   name: probes-images-v2.0
@@ -14,7 +14,7 @@
   selfLink: /apis/apps/v1/namespaces/default/replicasets/probes-images-v2.0
   uid: 1d717f4e-7410-11e9-9a36-eefc5b75fd0d
 spec:

-  replicas: 6
+  replicas: 2
   selector:
     matchLabels:
       app: probes-images
exit status 1
$ kubectl apply -f 508_probes-images-scale-down-both-rs.yaml && kubectl get pods -w -l app=probes-images
replicaset.apps/probes-images configured
replicaset.apps/probes-images-v2.0 configured
NAME                       READY   STATUS        RESTARTS   AGE
probes-images-62v48        1/1     Terminating   0          39m
probes-images-bs6z9        1/1     Terminating   0          39m
probes-images-ndhd7        1/1     Terminating   0          39m
probes-images-pbltv        1/1     Running       0          42m
probes-images-sv698        1/1     Running       0          42m
probes-images-v2.0-86shb   1/1     Running       0          28m
probes-images-v2.0-9kpln   1/1     Running       0          28m
probes-images-v2.0-dnl2k   1/1     Terminating   0          28m
probes-images-v2.0-dxj4v   1/1     Terminating   0          28m
probes-images-v2.0-hg8xb   1/1     Terminating   0          28m
probes-images-v2.0-mmvzr   1/1     Terminating   0          28m
probes-images-vbrqc        1/1     Terminating   0          42m
probes-images-v2.0-mmvzr   0/1     Terminating   0          28m
probes-images-62v48        0/1     Terminating   0          39m
probes-images-vbrqc        0/1     Terminating   0          42m
probes-images-bs6z9        0/1     Terminating   0          39m
probes-images-62v48        0/1     Terminating   0          39m
probes-images-v2.0-dnl2k   0/1     Terminating   0          28m
probes-images-v2.0-dnl2k   0/1     Terminating   0          28m
probes-images-vbrqc        0/1     Terminating   0          42m
probes-images-vbrqc        0/1     Terminating   0          42m
probes-images-vbrqc        0/1     Terminating   0          42m
probes-images-ndhd7        0/1     Terminating   0          39m
probes-images-ndhd7        0/1     Terminating   0          39m
probes-images-62v48        0/1     Terminating   0          39m
probes-images-v2.0-dxj4v   0/1     Terminating   0          28m
probes-images-62v48        0/1     Terminating   0          39m
probes-images-v2.0-hg8xb   0/1     Terminating   0          28m
probes-images-bs6z9        0/1     Terminating   0          39m
probes-images-bs6z9        0/1     Terminating   0          39m
probes-images-v2.0-mmvzr   0/1     Terminating   0          28m
probes-images-v2.0-mmvzr   0/1     Terminating   0          28m
probes-images-v2.0-hg8xb   0/1     Terminating   0          28m
probes-images-v2.0-hg8xb   0/1     Terminating   0          28m
$ kubectl get pods -l app=probes-images 
NAME                       READY   STATUS    RESTARTS   AGE
probes-images-pbltv        1/1     Running   0          42m
probes-images-sv698        1/1     Running   0          42m
probes-images-v2.0-86shb   1/1     Running   0          29m
probes-images-v2.0-9kpln   1/1     Running   0          29m
$  kubectl get pods -l app=probes-images -o custom-columns=NAME:.metadata.name,LABELS:.metadata.labels,OWNER-TYPE:.metadata.ownerReferences[0].kind,OWNER-NAME:.metadata.ownerReferences[0].name --sort-by=.status.startTime
NAME                       LABELS                                OWNER-TYPE   OWNER-NAME
probes-images-pbltv        map[app:probes-images version:v2.0]   ReplicaSet   probes-images
probes-images-sv698        map[app:probes-images version:v2.0]   ReplicaSet   probes-images
probes-images-v2.0-86shb   map[app:probes-images version:v2.0]   ReplicaSet   probes-images-v2.0
probes-images-v2.0-9kpln   map[app:probes-images version:v2.0]   ReplicaSet   probes-images-v2.0

And now scale both up!

$ kubectl diff -f 509_probes-images-scale-up-both-rs.yaml   
diff -u -N /tmp/LIVE-749110581/apps.v1.ReplicaSet.default.probes-images /tmp/MERGED-606903056/apps.v1.ReplicaSet.default.probes-images
--
- /tmp/LIVE-749110581/apps.v1.ReplicaSet.default.probes-images        2019-05-11 17:43:39.257897487 +0000
+++ /tmp/MERGED-606903056/apps.v1.ReplicaSet.default.probes-images      2019-05-11 17:43:39.278899367 +0000
@@ -5,7 +5,7 @@
     kubectl.kubernetes.io/last-applied-configuration: |
       {"apiVersion":"apps/v1","kind":"ReplicaSet","metadata":{"annotations":{},"labels":{"app":"probes-images"},"name":"probes-images","namespace":"default"},"spec":{"replicas":2,"selector":{"matchLabels":{"app":"probes-images"}},"template":{"metadata":{"labels":{"app":"probes-images","version":"v2.0"}},"spec":{"containers":[{"image":"raelga/cats:blanca","livenessProbe":{"failureThreshold":2,"httpGet":{"path":"/","port":80},"initialDelaySeconds":10,"periodSeconds":5},"name":"app","readinessProbe":{"httpGet":{"path":"/","port":80},"initialDelaySeconds":2}}]}}}}
   creationTimestamp: "2019-05-11T16:20:36Z"

-  generation: 24
+  generation: 25
   labels:
     app: probes-images
   name: probes-images
@@ -14,7 +14,7 @@
   selfLink: /apis/apps/v1/namespaces/default/replicasets/probes-images
   uid: b57546b3-7408-11e9-9a36-eefc5b75fd0d
 spec:

-  replicas: 2
+  replicas: 4
   selector:
     matchLabels:
       app: probes-images
diff -u -N /tmp/LIVE-749110581/apps.v1.ReplicaSet.default.probes-images-v2.0 /tmp/MERGED-606903056/apps.v1.ReplicaSet.default.probes-images-v2.0
--
- /tmp/LIVE-749110581/apps.v1.ReplicaSet.default.probes-images-v2.0   2019-05-11 17:43:39.420912073 +0000
+++ /tmp/MERGED-606903056/apps.v1.ReplicaSet.default.probes-images-v2.0 2019-05-11 17:43:39.440913863 +0000
@@ -5,7 +5,7 @@
     kubectl.kubernetes.io/last-applied-configuration: |
       {"apiVersion":"apps/v1","kind":"ReplicaSet","metadata":{"annotations":{},"labels":{"app":"probes-images"},"name":"probes-images-v2.0","namespace":"default"},"spec":{"replicas":2,"selector":{"matchLabels":{"app":"probes-images","version":"v2.0"}},"template":{"metadata":{"labels":{"app":"probes-images","version":"v2.0"}},"spec":{"containers":[{"image":"raelga/cats:blanca","livenessProbe":{"failureThreshold":2,"httpGet":{"path":"/","port":80},"initialDelaySeconds":10,"periodSeconds":5},"name":"app","readinessProbe":{"httpGet":{"path":"/","port":80},"initialDelaySeconds":2}}]}}}}
   creationTimestamp: "2019-05-11T17:13:37Z"

-  generation: 2
+  generation: 3
   labels:
     app: probes-images
   name: probes-images-v2.0
@@ -14,7 +14,7 @@
   selfLink: /apis/apps/v1/namespaces/default/replicasets/probes-images-v2.0
   uid: 1d717f4e-7410-11e9-9a36-eefc5b75fd0d
 spec:

-  replicas: 2
+  replicas: 4
   selector:
     matchLabels:
       app: probes-images
exit status 1
$ kubectl apply -f 509_probes-images-scale-up-both-rs.yaml && kubectl get rs -l app=probes-images -w 
replicaset.apps/probes-images configured
replicaset.apps/probes-images-v2.0 configured
NAME                 DESIRED   CURRENT   READY   AGE
probes-images        4         4         2       83m
probes-images-v2.0   4         4         2       30m
probes-images        4         4         3       83m
probes-images        4         4         4       83m
probes-images-v2.0   4         4         3       30m
probes-images-v2.0   4         4         4       30m
$  kubectl get pods -l app=probes-images -o custom-columns=NAME:.metadata.name,LABELS:.metadata.labels,OWNER-TYPE:.metadata.ownerReferences[0].kind,OWNER-NAME:.metadata.ownerReferences[0].name --sort-by=.status.startTime
NAME                       LABELS                                OWNER-TYPE   OWNER-NAME
probes-images-pbltv        map[app:probes-images version:v2.0]   ReplicaSet   probes-images
probes-images-sv698        map[app:probes-images version:v2.0]   ReplicaSet   probes-images
probes-images-v2.0-86shb   map[app:probes-images version:v2.0]   ReplicaSet   probes-images-v2.0
probes-images-v2.0-9kpln   map[app:probes-images version:v2.0]   ReplicaSet   probes-images-v2.0
probes-images-7qtn8        map[app:probes-images version:v2.0]   ReplicaSet   probes-images
probes-images-v2.0-n49jk   map[app:probes-images version:v2.0]   ReplicaSet   probes-images-v2.0
probes-images-v2.0-szbh4   map[app:probes-images version:v2.0]   ReplicaSet   probes-images-v2.0
probes-images-zmdfk        map[app:probes-images version:v2.0]   ReplicaSet   probes-images

It has become clear clear that ReplicaSet is not meant to be used for Rolling Updates or deploying new versions of our application. Is for keeping a number of replicas of a Pod running and that's all.

For manage application deployments, we need another kind of Kubernetes object that should work with ReplicaSets... Lucky for us, Kubernetes has an object for that called... you guessed... Deployments.

You can’t perform that action at this time.