Skip to content

Latest commit

 

History

History
219 lines (159 loc) · 7.64 KB

CONTRIBUTING.md

File metadata and controls

219 lines (159 loc) · 7.64 KB
Raw

Getting Started

These instructions will result in an (almost) fully function development environment for config-manager using minikube to run kubernetes and bonfire to deploy the dependent platform applications into the cluster.

Prerequisites

quay.io

Follow the instructions for generating a quay pull secret, but addition to downloading a pull secret file, copy the "podman login" instructions and log in.

You must have access to the quay.io/cloudservices repository. In your app-interface user file, make sure quay_username is set to your quay.io username and that you are a member of the insights-engineers.yml role.

minikube

Follow the instructions on installing minikube.

oc

Follow the instructions on installing oc.

bonfire

Follow the instructions on installing bonfire.

Set up a kubernetes cluster using minikube

You only need to do this once.

Start minikube

minikube runs as a libvirt domain. Setting the default libvirt connection URI and adding your user account to the libvirt UNIX group makes interacting with libvirt easier.

echo "LIBVIRT_DEFAULT_URI=qemu:////system" >> ~/.bashrc
sudo usermod -a -G libvirt $USER

Log out or reboot after adding yourself to the group to ensure the seat is completely logged out.

minikube start --cpus 8 --disk-size 36GB --memory 16GB --addons=registry --driver=kvm2

Install Clowder Custom Resource Definitions

curl https://raw.githubusercontent.com/RedHatInsights/clowder/$(curl https://api.github.com/repos/RedHatInsights/clowder/releases/latest | jq -r '.tag_name')/build/kube_setup.sh | bash

Install Clowder

minikube kubectl -- apply --filename $(curl https://api.github.com/repos/RedHatInsights/clowder/releases/latest | jq -r '.assets[0].browser_download_url')
# Apply the example clowder-config ConfigMap that sets enableExternalStrimzi to true
minikube kubectl -- --namespace clowder-system apply --filename https://raw.githubusercontent.com/RedHatInsights/clowder/$(curl https://api.github.com/repos/RedHatInsights/clowder/releases/latest | jq -r '.tag_name')/clowder-config.yaml
# Delete the clowder-system-controller-manager pod to rebuild with the above applied ConfigMap
minikube kubectl -- --namespace clowder-system delete pods --selector control-plane=controller-manager

Create a namespace

Choose a namespace to deploy applications into. The name you choose doesn't matter, but you'll be typing it a lot. Pick something short. I like 'fog' (because fog is a cloud on the ground... get it?).

minikube kubectl -- create namespace fog

Create Pull Secrets

Some deployments are hard-coded to use a secret named quay-cloudservices-pull, so copy the pull secret you downloaded above and rename it:

cp $USER-secret.yml quay-cloudservices-pull.yml
sed -ie "s/$USER-pull-secret/quay-cloudservices-pull/" quay-cloudservices-pull.yml

Now create the secrets in your new namespace:

minikube kubectl -- --namespace fog create --filename $USER-secret.yml
minikube kubectl -- --namespace fog create --filename quay-cloudservices-pull.yml

Deploy ClowdEnvironment

bonfire deploy-env --namespace fog --quay-user $USER

Deploy Applications

bonfire deploy --namespace fog --local-config-path ./scripts/bonfire.yml \
    cloud-connector \
    host-inventory \
    playbook-dispatcher

Run config-manager

There are two ways to run config-manager: directly on localhost or deployed into the minikube cluster. Both have advantages and drawbacks, so both are presented equally here.

Run directly on localhost

Running directly on localhost allows for a more rapid edit/execute/debug development cycle, but requires forwarding ports from localhost into the cluster services.

Forward ports

The included kube-port-forward.sh script will forward ports from localhost to all the necessary services in the cluster.

bash scripts/kube-port-forward.sh
service local port cluster port
mosquitto 1883 1883
playbook-dispatcher-api 8001 8000
host-inventory-service 8002 8000
cloud-connector 8003 8080
kafka-ext-bootstrap 9094 9094

Run database

podman run --env POSTGRES_PASSWORD=insights --env POSTGRES_USER=insights --env POSTGRES_DB=insights --publish 5432:5432 --name config-manager-db --detach postgres

Run config-manager

CM_LOG_LEVEL=debug \
CM_KAFKA_BROKERS=localhost:9094 \
CM_DISPATCHER_HOST=http://localhost:8001/ \
CM_DISPATCHER_PSK=$(kubectl -n fog get secrets/psk-playbook-dispatcher -o json | jq '.data.key' -r | base64 -d) \
CM_INVENTORY_HOST=http://localhost:8002/ \
CM_CLOUD_CONNECTOR_HOST=http://localhost:8003/api/cloud-connector/v1/ \
CM_CLOUD_CONNECTOR_PSK=$(kubectl -n fog get secrets/psk-cloud-connector -o json | jq '.data["client-psk"]' -r | base64 -d) \
CM_WEB_PORT=8080 \
CM_DB_USER=insights \
CM_DB_PASS=insights \
CM_DB_NAME=insights \
go run .

Deployed into the cluster

Running config-manager in the cluster more accurately simulates the environment in which config-manager runs in the production and stage environments. It more seamlessly interacts with other applications (such as cloud-connector and playbook-dispatcher), but makes for a much slower development cycle. Every time a code change is made, a new image has to be built, pushed and deployed. Fortunately, there is a convenient script to do this for you. See scripts/README.md for details.

bash scripts/bonfire-deploy.sh

Forward the port

The included kube-port-forward.sh script will forward ports from localhost to all the necessary services in the cluster, including the config-manager-service if detected.

bash scripts/kube-port-forward.sh

Send HTTP requests

It should now be possible to interact with config-manager's HTTP API using curl or ht.

ht GET http://localhost:8080/api/config-manager/v1/states/current x-rh-identity:$(xrhidgen user | base64 -w0)

Debugging

Monitoring Kafka topics

# Identify the environment name and export it
export CONFIG_MANAGER_ENV=$(kubectl -n fog get svc -l env=env-fog,app.kubernetes.io/name=kafka -o json | jq '.items[0].metadata.labels["app.kubernetes.io/instance"]' -r)
minikube kubectl -- -n fog run -it --rm --image=edenhill/kcat:1.7.1 kcat -- -b $CONFIG_MANAGER_ENV-kafka-bootstrap.fog.svc.cluster.local:9092 -t platform.inventory.events

Produce an Inventory Event Kafka message

# Identify the environment name and export it
export CONFIG_MANAGER_ENV=$(kubectl -n fog get svc -l env=env-fog,app.kubernetes.io/name=kafka -o json | jq '.items[0].metadata.labels["app.kubernetes.io/instance"]' -r)
jq --compact-output --null-input --arg id $(uuidgen | tr -d "\n") '{"type":"created","host":{"id":$id,"account":"0000001","reporter":"cloud-connector","system_profile":{"rhc_client_id":$id}}}' | minikube kubectl -- -n fog run -i --rm --image=edenhill/kcat:1.7.1 $(mktemp XXXXXX) -- -b $CONFIG_MANAGER_ENV-kafka-bootstrap.fog.svc.cluster.local:9092 -t platform.inventory.events -P -H event_type=created