Skip to main content

Deploying using Helm

This guide helps you setup a FlowG Kubernetes cluster using our Helm charts.

This guide splits into 3 different branches, following the needs of your cluster.

  1. Single-node FlowG instance with BadgerDB as a backend, if you only need a small installation.
  2. Multi-node cluster using foundationDB (and optionnally an observability stack), if you need multiple instances running

Prerequisites​

  • A working Kubernetes cluster
  • Helm

To size the pods (flowg.resources, flowg.goMemLimit) for your log throughput, see Tuning.

1. Single-node FlowG instance​

First, clone the repository containing our helm charts:

git clone https://github.com/link-society/flowg
cd flowg/k8s/charts

Configuration
The configuration can be found inside values.yaml. The default configuration normally applies to all needs, but you can still customize it if you want. See the configuration reference.

To install the chart, simply run:

helm install flowg flowg-singlenode -n flowg-system --create-namespace

This will create these resources into the chosen namespace:

2. Multi-node cluster, with/out Fluentd​

Installing FoundationDB CRD​

In order to make FlowG work with FoundationDB, you first need to intall the FoundationDB Kubernetes Operator.

To do so, a simple way is to use the Helm chart they provide.

# Clone the repository 
git clone https://github.com/foundationdb/fdb-kubernetes-operator
cd fdb-kubernetes-operator

# Install FoundationDB CRD
# We install it in the same namespace as FlowG 
helm install foundationdb charts/fdb-opeator -n flowg-system --create-namespace

Setting up FlowG​

The instructions are close to the ones for the other charts. First, you need to clone the repository:

git clone https://github.com/link-society/flowg
cd flowg

Configuration
The configuration can be found inside values.yaml. The default configuration normally applies to all needs, but you can still customize it if you want. See the configuration reference.

And then install the chart following your needs:

  • If you don't need fluentD:
helm install flowg flowg-multinode -n flowg-system --create-namespace
  • If you want an observability stack based on fluentd:
helm install flowg flowg-observability -n flowg-system --create-namespace

Fluentd
Fluentd manages all the log routing to FlowG by gathering ALL the logfiles specified in the logdir in values.yaml, except fluentd's own logs, and sends it to FlowG's syslog server.
The applied configuration is minimal and only parse the entries following the CRI (Kubernetes) format to send it under Json format to FlowG.

Important note
The cluster will take some time to get brought up. This is normal. FoundationDB takes time reconciling their cluster and creating the cluster file. An init container waiting for it will be waiting for it to exist before running FlowG (to avoid it crashing on startup).

info

To get a list of every available features in each chart, please refer to the full configuration reference : Technical Documentation / Helm