Skip to content

Latest commit

 

History

History
129 lines (93 loc) · 2.78 KB

readme.asciidoc

File metadata and controls

129 lines (93 loc) · 2.78 KB
Raw

Side By Side Demo

Offers Kubernetes Deployments and Knative Serving services of:

  • dotnetcore

  • Go

  • Nodejs

  • Python

  • Quarkus

  • Spring Boot

Quickstart with Knative Serving

kubectl create namespace sidebyside
kubens sidebyside

OR

oc new-project sidebyside

Node.js

cd hello/nodejs
kubectl apply -f kubefiles/knService_docker.yml
# the last I checked, Knative Serving worked best with docker hub due to registry API v2_2
./knpoller_ocp4.sh
# there other poller scripts depending on your Kubernetes variant
Node Hello on noded-wj854-deployment-569f8cd6cd-vn9g9:1
Node Hello on noded-wj854-deployment-569f8cd6cd-vn9g9:2

Quarkus

cd hello/quarked
kubectl apply -f kubefiles/knService_docker.yml
./knpoller_ocp4.sh
Aloha quarked-qwt7d-deployment-6f4ccfffc5-g2xxg:1
Aloha quarked-qwt7d-deployment-6f4ccfffc5-g2xxg:2

Spring Boot

cd hello/booted
kubectl apply -f kubefiles/knService_docker.yml
./knpoller_ocp4.sh
Hola Boot booted-kskv5-deployment-68f8f944ff-h8lvs:1
Hola Boot booted-kskv5-deployment-68f8f944ff-h8lvs:2

Burst

The Knative Service yaml definition includes the following:

        metadata:
          annotations:
            autoscaling.knative.dev/target: "10"

The out-of-the-box default is 100 which is hard to "overload" and cause more than 1 pod to spawn.

The knBurst scripts use a tool called siege to send in 40 concurrent requests, causing Knative to scale to 4 pods.

Terminal 1:

watch kubectl get pods

Terminal 2:

./knburst_ocp4.sh
** SIEGE 4.0.4
** Preparing 40 concurrent users for battle.
The server is now under siege..

Terminal 1:

NAME                                      READY   STATUS              RESTARTS   AGE
noded-wj854-deployment-569f8cd6cd-9bwtl   1/2     Running             0          11s
noded-wj854-deployment-569f8cd6cd-db7vl   1/2     Running             0          10s
noded-wj854-deployment-569f8cd6cd-j4bsm   0/2     ContainerCreating   0          8s
noded-wj854-deployment-569f8cd6cd-kxr2f   1/2     Running             0          10s

Dashboards

Prometheus

open https://$(oc -n openshift-monitoring get route prometheus-k8s -o jsonpath='{..spec.host}')

Or else you can open the integrated prometheus dashboard within OpenShift:

openshift menu

Metrics

The PromQL is

container_memory_rss{namespace="sidebyside",container="user-container"}

You should then see a graph like this one:

prometheus memory query

Grafana

open https://$(oc -n openshift-monitoring get route grafana -o jsonpath='{..spec.host}')

Select Kubernetes / Compute Resources / Namespace (Pods)

Kubernetes / Compute Resources / Namespace (Pods)