Flair Helm charts

Helm charts in this repository help you deploy Flair in Kubernetes environment. This documemtation takes you though a step-by-step guide to deploying all the services in Kubernetes

Note: The following instructions have been tested on a GKE cluster and on minikube

Helm Repo

You can add Flair Platform chart repo to and start deploying services.

helm repo add flair-platform https://flair-platform.storage.googleapis.com

If you already have the repo added, consider updating your local index by running

helm repo update

Listing charts

You can list for all the charts available in the flair-platform repo by running the following command

helm search repo flair-platform

Alternatively, you can follow the steps below, if you wish you deploy charts from the code base itself.

Prerequisites

• Minikube
• Kubectl
• Helm
• Docker cli

Starting Minikube

Once yo have minikube installed, you can go ahead and start a minikube version of Kubernetes.

Note that we are giving minikube 5 CPUs(vCPUs) and 10 Gi of RAM. The verion of Kubernetes we are usign is 1.14.2

minikube start --cpus 5 --memory 10240 --kubernetes-version=v1.14.2 --profile flair

Once minikube is up and running, please go ahead and initialize the tiller pod necessary for helm

helm init

This might take a few minuites for the tiller pod to be available to take any command.

If you are looking to build docker images from flair's source repositories, you would have to run the following command to point to minikubes docker demon

eval $(minikube docker-env)

Note: The recommonded way however is to use Vizcentric published docker images in Docker registry.

Deploying Flair Services

Since we now have the prerequisites necessary for us to deploy flair services sorted, we can go ahead and deploy the services in Kubernetes. Please stick to the order as mentioned below.

Local Repository and environment setup

Clone this repository and cd into the folder

git clone https://github.com/viz-centric/flair-helm-charts.git
cd flair-helm-charts

You can follow along with these step-by-step installation or optionally you can also install all the services through the setup.sh script which does everything for you. You can then skip the following steps and go onto Accessing Flair in Kubernetes section.

./setup.sh

If not, you can follow on with the next steps

Create a namespace into which we are going to deploy all Flair services

kubectl create namespace flair

Note: We will be using this namespace to deploy all our services into. It would be easy to use kubens flair so that your default namespace is changed to flair and simplifies yoour commands.

Flair Registry

Flair Registry is used for distributing configuration to all the services and hence the first service which goes in. You can install the flair registry chart by running the following command.

helm upgrade \
    --install \
    --wait \
    --namespace flair \
    flair-registry ./flair-registry

Flair Engine

Once you have Flair Registry up and running we can go ahead with deploying flair engine. For us to do so, we would have to deploy Postgres first as it is the state store for this service.

Postgres

helm upgrade \
    --install \
    --wait \
    --namespace flair \
    --values ./flair-postgres/flair-engine.yaml \
    flair-engine-pg stable/postgresql

Note: When deploying Flair in production, its recommonded to run postgres outside kubernetes. For example in Public clouds you can make use of cloud provideres offering for Postgres 9.4+. Make sure you make changes to the configuration to point to postgres accordingly.

You can now deploy Flair Engine by running the following command.

helm upgrade \
    --install \
    --wait \
    --namespace flair \
    flair-engine ./flair-engine

Flair Cache

You can now deploy Flair Cache

helm upgrade \
    --install \
    --wait \
    --namespace flair \
    flair-cache ./flair-cache

Flair Notifications

Once you have Flair Cache up and running we can go ahead with deploying flair notifications service. For us to do so, we would have to deploy Postgres first as it is the state store for this service.

Postgres

helm upgrade \
    --install \
    --wait \
    --namespace flair \
    --values ./flair-postgres/flair-notifications.yaml \
    flair-notifications-pg stable/postgresql

Note: When deploying Flair in production, its recommonded to run postgres outside kubernetes. For example in Public clouds you can make use of cloud provideres offering for Postgres 9.4+. Make sure you make changes to the configuration to point to postgres accordingly.

You can now deploy Flair Notifications by running the following command.

helm upgrade \
    --install \
    --wait \
    --namespace flair \
    flair-notifications ./flair-notifications

Flair BI

For you to install flair bi you will need a postgres database and CouchDB where data is persisted. We can start by installing both of those first

Postgres

helm upgrade \
    --install \
    --wait \
    --namespace flair \
    --values ./flair-postgres/flair-bi.yaml \
    flair-bi-pg stable/postgresql

Note: When deploying Flair in production, its recommonded to run postgres outside kubernetes. For example in Public clouds you can make use of cloud provideres offering for Postgres 9.4+. Make sure you make changes to the configuration to point to postgres accordingly.

CouchDB

helm upgrade \
    --install \
    --wait \
    --namespace flair \
    --values ./flair-couchdb/values.yaml \
    flair-couchdb ./flair-couchdb/couchdb

Once postgres and couchdb are sucessfully deployed, go ahead and deploy flair bi

helm upgrade \
    --install \
    --wait \
    --namespace flair flair-bi ./flair-bi

Flair with ISTIO Service-Mesh

Get Istio

curl -L https://git.io/getLatestIstio | ISTIO_VERSION=1.3.1 sh -
cd istio-1.3.1
export PATH=$PWD/bin:$PATH

Install CRDs

for i in install/kubernetes/helm/istio-init/files/crd*yaml; do kubectl apply -f $i; done

Setup Istio

# For permissive multual TLS
kubectl apply -f install/kubernetes/istio-demo.yaml

Enable mTLS and deploy Flair

./secure-mesh.sh
# Wait until all services are up and running
# aws-athena-query-results-689779572241-eu-west-2

Sniffing traffic from FLAIRBI to FLAIRENGINE

kubectl exec -it <ANY_FLAIRBI_POD> -c istio-proxy
ifconfig
sudo tcpdump -vvvv -A -nn -i eth0 '(dst port 6565)'

Accessing Flair in Kubernetes

Once all the services are up and running you can access Flair BI to start visualizing data.

In Minikube

You can now port forward the Flair BI's service onto to your machine by running the following command

kubectl port-forward svc/flair-bi 8002:8002

In production

In production we would recommand using public cloud providers load balancer annotation to provision a load balancer for Flair BI alone and create an Alias record (A) in the DNS service pointing at the load balancer. Flair will then be available for your users though the configured DNS name.

Clean UP

You can delete all the services by running the clean_up.sh script to delete all flair services and remove the created namespace.

./clean_up.sh

or

You can manually delete all the services in the cluster by running the following commands

# Removing all services
helm delete flair-bi --purge
helm delete flair-bi-pg --purge
helm delete flair-couchdb --purge
helm delete flair-notifications-pg --purge
helm delete flair-notifications --purge
helm delete flair-cache --purge
helm delete flair-engine --purge
helm delete flair-engine-pg --purge
helm delete flair-registry --purge
# Removing the namesapce
kubectl delete namespace flair

If you have ran all of this in minikube, you can also delete the minikube vm by running

minikube stop --profile flair
minikube delete --profile flair

Roadmap

New Services

• Flair Operator

Note: If you are planning to deploy flair in any other way, please let us know and we can help you out developing and deploying it.