infra - GitLite

KO | EN

gitlite — search

#javascript #hacktoberfest #react #typescript #android #rust #nodejs #java #cli #kubernetes #pytorch #ios #openai #open-source

infra

(★ 109)

DEPRECATED: Use as reference but better use kubeadm/bootkube for cluster setup!

파일 탐색기

최종 버전 다운로드 (.zip)

- - ca-config.json
  - ca-csr.json
- env
- base.json
- install.sh
- Makefile
- README.md
- configure.sh
- id_rsa.pub
- main.tf
- README.md
- terraform
- upgrade
.gitignore
mk_credentials
README.md

# 설치 가이드

1. 코드 내려받기

git clone https://github.com/5pi/infra

깃허브에서 프로젝트 코드 전체를 내 컴퓨터로 내려받습니다.

cd infra

방금 내려받은 프로젝트 폴더 안으로 이동합니다.

2. Make

보통 추천

사전 준비물

Git GitHub에서 프로젝트 코드를 내려받으려면 필요합니다.
Make Linux/macOS는 보통 기본 설치되어 있습니다. Windows는 별도 설치(예: MSYS2, WSL)가 필요합니다.

cd packer

이 프로젝트의 관련 파일이 하위 폴더 안에 있어서, 먼저 그 폴더로 이동합니다.

make

생성된 빌드 설정을 바탕으로 실제 컴파일을 진행해 실행 파일을 만듭니다.

✅ 에러 없이 끝나면 성공입니다. 생성된 실행 파일을 직접 실행해보세요.

// repository documentation

Was this content helpful?

★ 0 (0 ratings)

# 5pi.de Infrastructure
*$15/month kubernetes cluster*

This is how I deploy Kubernetes to DigitalOcean.
This aims to be generic enough to also deploy your infrastructure but there
might be some specifics. The biggest limitation right now is that all servers
are both master and minion which is discouraged for any larger infrastructure.

While it's possible that this will become a generic way to setup any kind of
Kubernetes infrastructures, it's not a top priority. If interested in a
deployment on AWS, have a look at [kops](https://github.com/kubernetes/kops)
which is more active developed.

## Overview
Terraform deploys the infrastructure by setting up network, spinning up servers
with an image build by Packer.

This repository contains [packer](/packer) and [terraform](/tf) configuration.
The directory [config](/config) contains configuration that is shared between
the image and terraform templates. See the individual directories for details.

The infrastructure is designed to be immutable. All state is intended to be kept
on DigitalOcean volumes and all change to the host require a new image and
replacement of all instances.

## Deploying a new stack
First you might want to edit `config/env` to customize:

- `REGION`: The DigitalOcean region your cluster should run in
- `DOMAIN`: The Domain used for the cluster (see [tf](/tf) for details)
- `SERVERS`: Number of servers
- `SERVER_SIZE`: Size of servers
- `IP_INT_PREFIX`: Prefix to use for internal private network (tinc)
- `CA_FILE`: The full path (on the servers) for your TLS CA cert file

You might also want to edit `config/ca/ca-csr.json`.

Then run `mk_credentials` to create TLS CA and keys in `config/generated/`.

Now the image can be build by running `make -C packer`. After finishing, it
should print image id which is used in the next step.

Enter the `tf/` directory and save a ssh public key to `id_rsa.pub`. This key
will be allowed to ssh into the servers.

A DigitalOcean API token is required for running Packer and Terraform as well as
for attaching the DigitalOcean Volumes. This scripts expect it in `~/.do-token`.

Now the stack can be spun up by running:

```
./terraform apply -var cluster_state=new -var image=image-id-from-last-step
```

## Updating a stack
Since the servers are immutable, configuration shouldn't be changed on the
systems directly. Instead a new image should be built. Once the build finished,
the stack can be updated by running this in `tf/`:

```
./upgrade apply -var image=image-id-from-build
```

This is a small wrapper around terraform to apply changes to the cluster one
server at a time. It removes a server from the cluster gracefully and waits for
a replacement to come up and successfully join the cluster.