# Dependencies and Pre-requisites

Pipekit depends on the following tools being available in your environment:

## Kubernetes

* We recommend Kubernetes 1.33 or greater.

### Setup

* The installation makes use of ingresses to route traffic to the various services. You will need to have an ingress controller installed in your cluster. We recommend using the [ingress-nginx](https://kubernetes.github.io/ingress-nginx/) controller.
* OTEL or Prometheus will allow you to monitor the pipekit services on their defined metrics ports. This is not a strict requirement but is recommended.
* An in-cluster log collection tool such as Fluentd and an aggregation tool (e.g. Loki) is recommended to collect and display logs from the various services. This is not a strict requirement but is recommended.
* The Pipekit services make use of HPAs to scale under load. You will need to have the metrics server installed in your cluster in order to make use of this.

## Postgres

* We recommend you use a managed Postgres service such as AWS RDS or Azure Postgres.
* We recommend Postgres 18 or greater.

### Setup

* The Pipekit services require a Postgres database to store their data. You will need to create a database and user for the `id` and `users` services to use:

  ```bash
  CREATE DATABASE users;
  CREATE DATABASE id;
  ```

## Redis

* Can be inside the cluster or a managed service such as AWS Elasticache. Internally, we use Valkey as a Redis compatible database.
* We recommend Redis or Valkey 8 or greater running in a high availability setup with sentinel.
* We recommend configuring disk persistence for Redis. However, there is no need to back up this disk. If the data is lost, Pipekit will recreate it, but in-flight workflows may not have the correct state in Pipekit if Redis' data is lost at the time the workflow is being processed.

### Setup

No special configuration is required for Redis. We would recommend that it is configured to be highly available. It would also be advisable to configure Redis to use a password which can be passed to the Pipekit Helm Chart.

## Grafana Loki

* Can be hosted within the Kubernetes cluster, or externally.
* We recommend Loki 3.5 or greater.

### Setup

No special configuration is required for Loki. We would recommend that it is configured to be highly available.

## Git Provider (GitHub, GitLab)

You can optionally connect your Git provider to Pipekit in order to make use of the GitOps features. While this is optional, it is highly recommended.

### Setup

Follow the instructions in the [Git Provider](https://docs.pipekit.io/self-hosting-pipekit/integrating-with-your-git-provider) documentation to connect your Git provider to Pipekit.