This is a demo app to showcase orchestrating Docker with Kubernetes. You can find more at my post.
- Run backend
kubectl create -f backend.yaml
According to the best practices, create a Service before corresponding Deployments so that the scheduler can spread the pods comprising the Service. So the redis service is created before deployment and they are defined in backend.yaml.
- Run frontend
kubectl create -f frontend.yaml
Note I used type: NodePort, because our cloud doe not provide public IP.
The solution is to map the node port to container port,
then in LB configure virtual port to node port.
For example:
http 30003 http 30003
After that one can find the FQDN for the node and use below to see the page in browser.
http://<fqdn-for-node-lb>:30003/
Note both should use LB and FQDN for Node, Not Master.
- Self-healing
Delete a random pod and see Kubernetes creates a new one.
kubectl -s http://<fqdn-for-master-lb>:8080 delete pod frontend-1287392616-6kk6q; kubectl -s http://<fqdn-for-master-lb>:8080 get pods | grep frontend
pod "frontend-1287392616-6kk6q" deleted
frontend-1287392616-3zmqa 1/1 Running 0 27m
frontend-1287392616-6kk6q 1/1 Terminating 0 27m
frontend-1287392616-8lki8 0/1 Terminating 0 1s
frontend-1287392616-92fgg 1/1 Running 0 27m
frontend-1287392616-my2gy 0/1 Terminating 0 1s
frontend-1287392616-r9sfx 0/1 Pending 0 1s
kubectl -s http://<fqdn-for-master-lb>:8080 get pods | grep frontend
frontend-1287392616-3zmqa 1/1 Running 0 28m
frontend-1287392616-92fgg 1/1 Running 0 28m
frontend-1287392616-r9sfx 1/1 Running 0 29s
- Scaling
kubectl -s http://<fqdn-for-master-lb>:8080 scale deployment/frontend --replicas=6; kubectl -s http://<fqdn-for-master-lb>:8080 get pods | grep frontend
deployment "frontend" scaled
frontend-1287392616-3zmqa 1/1 Running 0 32m
frontend-1287392616-7dd1z 0/1 Terminating 0 0s
frontend-1287392616-92fgg 1/1 Running 0 32m
frontend-1287392616-p1x6m 0/1 ContainerCreating 0 0s
frontend-1287392616-r9sfx 1/1 Running 0 4m
frontend-1287392616-z7jjv 0/1 ContainerCreating 0 0s
kubectl -s http://<fqdn-for-master-lb>:8080 get pods | grep frontend
frontend-1287392616-3zmqa 1/1 Running 0 32m
frontend-1287392616-92fgg 1/1 Running 0 32m
frontend-1287392616-p1x6m 1/1 Running 0 8s
frontend-1287392616-pjf35 1/1 Running 0 8s
frontend-1287392616-r9sfx 1/1 Running 0 4m
frontend-1287392616-z7jjv 1/1 Running 0 8s
kubectl -s http://<fqdn-for-master-lb>:8080 scale deployment/frontend --replicas=3; kubectl -s http://<fqdn-for-master-lb>:8080 get pods | grep frontend
deployment "frontend" scaled
frontend-1287392616-3zmqa 1/1 Running 0 33m
frontend-1287392616-92fgg 1/1 Running 0 33m
frontend-1287392616-lnuj6 0/1 Pending 0 0s
frontend-1287392616-p1x6m 1/1 Terminating 0 1m
frontend-1287392616-pjf35 1/1 Terminating 0 1m
frontend-1287392616-r9sfx 1/1 Running 0 6m
frontend-1287392616-z7jjv 1/1 Terminating 0 1m
kubectl -s http://<fqdn-for-master-lb>:8080 get pods | grep frontend
frontend-1287392616-3zmqa 1/1 Running 0 33m
frontend-1287392616-92fgg 1/1 Running 0 33m
frontend-1287392616-lnuj6 1/1 Terminating 0 5s
frontend-1287392616-r9sfx 1/1 Running 0 6m
frontend-1287392616-u8mjt 1/1 Terminating 0 4s
kubectl -s http://<fqdn-for-master-lb>:8080 get pods | grep frontend
frontend-1287392616-3zmqa 1/1 Running 0 36m
frontend-1287392616-92fgg 1/1 Running 0 36m
frontend-1287392616-r9sfx 1/1 Running 0 8m
- Rolling update
kubectl -s http://<fqdn-for-master-lb>:8080 create -f frontend-canary.yaml
kubectl -s http://<fqdn-for-master-lb>:8080 get pods -L track
NAME READY STATUS RESTARTS AGE TRACK
backend-3180978658-93bi0 1/1 Running 0 50m <none>
frontend-1287392616-3zmqa 1/1 Running 0 39m stable
frontend-1287392616-92fgg 1/1 Running 0 39m stable
frontend-1287392616-r9sfx 1/1 Running 0 11m stable
frontend-canary-1722551660-obhmi 1/1 Running 0 18s canary
kubectl -s http://<fqdn-for-master-lb>:8080 get deployments frontend-canary
NAME DESIRED CURRENT UP-TO-DATE AVAILABLE AGE
frontend-canary 1 1 1 1 2m
kubectl create -f frontend-canary.yaml
vim frontend.yaml # update simple-node:v1 to simple-node:v2
kubectl apply -f frontend.yaml
Hit the frontend service a few times to verify that only one pod is updated to chenglong/simple-node:v2
$ curl http://<fqdn-for-node-lb>:30003
This request is served by frontend-canary-1722551660-obhmi
You have viewed this page 1877 times!
Server Time: 2017-03-28T17:46:43.299Z
$ curl http://<fqdn-for-node-lb>:30003
This request is served by frontend-1287392616-r9sfx. You have viewed this page 1930 times!
Since the Canary Release is working fine, I want to roll out to all pods.
vim frontend.yaml
#image: chenglong/simple-node:v1
image: chenglong/simple-node:v2
kubectl -s http://<fqdn-for-master-lb>:8080 apply -f frontend.yaml
Find out details of one of the none-canary pods. Note that the image is chenglong/simple-node:v2 not v1.
kubectl -s http://<fqdn-for-master-lb>:8080 describe pods frontend-1389432169-t89d8
Name: frontend-1389432169-t89d8
Namespace: default
...
Containers:
nodejs:
Container ID: docker://9f3aded8f61cf6326ffcb4aa02699b4fd11ef713b0e879237e31040b79c0b60f
Image: chenglong/simple-node:v2
...
Delete Canary deployment
kubectl -s http://<fqdn-for-master-lb>:8080 delete deployments frontend-canary
If this release is not ideal, we could easily roll back
kubectl -s http://<fqdn-for-master-lb>:8080 rollout undo deployment/frontend
Confirm it is using v1 by describing one of the frontend pods.
kubectl -s http://<fqdn-for-master-lb>:8080 describe pods frontend-1287392616-2kakj
Name: frontend-1287392616-2kakj
...
Containers:
nodejs:
Container ID: docker://4373cb8ae9f8ac78f715e3a1a35ca652a8c20eb7452f92731a648860cd44a142
Image: chenglong/simple-node:v1
...