Profiling a service

Take alook at https://wikitech.wikimedia.org/wiki/User:Alexandros_Kosiaris/Benchmarking_kubernetes_apps

• Draft a new deployment-charts from service's template

  ./create_new_service.sh
  

• Get latest service's image
• Write locust.io scripts to test endpoints
• Collect data and tweak memory and CPU

minikube start
minikube addons enable metrics-server

Helm package

cd /path/to/charts/folder
helm package geoshapes
helm install geoshapes ./geoshapes-0.0.1.tgz

Uninstall package

During development you don't need to keep track of deployments, you just need to uninstall de package and reinstall:

helm uninstall geoshapes && helm package geoshapes && helm install geoshapes geoshapes-0.0.1.tgz

Expose the service ports to the local environment

kubectl get pods
kubectl port-forward <pod-name> 6534:6534

Open k8s dashboard

 minikube dashboard

Create Docker image

make build

Run Profiler

make run SERVICE_NAME=<service name> SERVICE_PORT=<service port>

• note: service name must match service names in service directory
• another note: service port comes from the service URL

Adding secrets to your service

Inspired by https://cloud.google.com/kubernetes-engine/docs/tutorials/authenticating-to-cloud-platform

kubectl create secret generic fcm-key --from-file=key.json=/home/mateus/workspace/wikimedia/infrastructure/fir-test-c2760-firebase-adminsdk-jachy-2cb00aa806.json

When the pod fail with errors

The following error indicates that the VM is not proper allocating resources, and the only solution I have so far is recreating the environment: 0/1 nodes are available: 1 Insufficient cpu, 1 Insufficient memory.

Solution:

minikube stop && minikube delete && minikube start && helm init

Debug helm templates

Take a look at this stackoverflow question: https://stackoverflow.com/questions/59564379/how-to-debug-helm-chart-errors-like-error-converting-yaml-to-json-yaml-mappin