gotodo

Getting started

touch .env
make start

Docker

make docker-update

This command will ask for loging, bump the version, build the new docker image and push it to the private repo.

Kubernetes

# If `Error: could not find tiller` run:
kubectl -n kube-system delete deployment tiller-deploy
kubectl -n kube-system delete service/tiller-deploy

WIP

• v2 with redis
• docker-compose vs Kompose
• Create docker image
• Add docker image to registry
• Init Helm in minikube
• Create Chart
• Check CRUD locally
• Create helm release
• Read about Makefile
• Fix Makefile
• Replace Minikube with docker-for-desktop
• Replace cloud.canister.io with registry.gitlab.com
• Command for

kubectl create secret generic regcred \ --from-file=.dockerconfigjson=<path/to/.docker/config.json> \ --type=kubernetes.io/dockerconfigjson

• Find to way to get VERSION to helm Values because tag ":latest" doesn't work
• Change to docker login --password-stdi
• Add project to CircleCI
• Read about CircleCI and Kubernetes
• Do we have to use a cloud provider to use CircleCI with Kubernetes?
• Create gcp cluster
• Make sure the db (Redis) tests are passing
• Remove everything from charts/values, secret values should be in .env and the rest in the Makefile
• Change selectors
• Change docker image registry. Binary will be copied by CirclCI
• Connect to gcp-cluster by edit ~/.kube/config file
• Setup k8s loadbalancer in helm chart and run curl localhost:3001/todos
• Run pod kubectl exec -it gotodo-b6487675f-xtg2q /bin/sh with command: ["/bin/sh"], tty: true and stdin: true
• Is Chart.yaml required?
• Add Jenkins and/or CircleCI
• Fix CHART_NAME
• Deploy on GCP
• Run all tests (including the one with Redis) on CircleCI for a specific branch
• Replace NodePort by LoadBalancer
• Deploy manually gotodo with helm
• Add tests
• Read more about Ingress
• Replace LoadBalancer with Ingress
• Debug go code inside docker container
• Write some doc about it
• Find equivalent to package.json/requirements.txt for go
• Check auth
• Oauth with google
• session in Redis
• Update memory profile commands
• Add simple gotodo frontend
• Organise backend, frontend code and devops code
• Update frontend with tailwindcss

Coverage report

go test -coverprofile temp/cover.out ./... && go tool cover -html=temp/cover.out

Memory profile

go test -memprofilerate 1 -memprofile temp/mem.out ./model
# Get model from ...
go tool pprof -web temp/model.a temp/mem.out

TODO: Shut idle pods down

source

#!/bin/bash
set -eo pipefail

ingress="$(kubectl get pods --output=jsonpath='{.items[*].metadata.name}' |
  xargs -n1 | grep "ingress-nginx" | head -n1)"

# cache all hosts that pass through the ingress
hosts="$(kubectl get ingress nginx-ingress \
  --output=jsonpath='{.spec.rules[*].host}' | xargs -n1)"

# cache pods
pods="$(kubectl get pods)"

# cache ingress logs of the last 90min
logs="$(kubectl logs --since=90m "$ingress")"

# iterate over all deployments
kubectl get deployment --output=jsonpath='{.items[*].metadata.name}' |
  xargs -n1 |
  while read -r svc; do

    # skip svc that don't have pods running
    echo "$pods" | grep -q "$svc" || {
      echo "$svc: no pods running"
      continue
    }

    # skip svcs that don't pass through the ingress
    echo "$hosts" | grep -q "$svc" ||  {
      echo "$svc: not passing through ingress"
      continue
    }

    # skip svcs with pods running less than 1h
    echo "$pods" | grep "$svc" | awk '{print $5}' | grep -q h ||  {
      echo "$svc: pod running less than 1h"
      continue
    }

    # check if any traffic to that svc was made through the ingress in the
    # last hour, scale it down case none
    echo "$logs" | grep -q "default-$svc" || {
      echo "$svc: scaling down"
      kubectl scale deployments "$svc" --replicas 0 --record || true
    }
  done