IPFS Infrastructure

Tools for maintaining infrastructure for the IPFS community.

• Introduction
• Getting started
• Usage
• Known issues
• Common tasks

Progress and Tracking

Introduction

This repository contains the technical infrastructure of the IPFS community.

• Public HTTP-to-IPFS Gateway: https://ipfs.io
• Default bootstrap used by IPFS: ipfs bootstrap
• Private networking between the hosts (cjdns with nginx access control)
• Monitoring of services and hosts: http://metrics.ipfs.team
• Pinbot, an IRC bot in chat.freenode.net/#ipfs-pinbot

Infrastructure that isn't contained here:

• Websites deployment: ipfs.io, dist.ipfs.io, blog.ipfs.io, chat.ipfs.io
• DNS settings for ipfs.io, ipld.io, multiformats.io, libp2p.io, orbit.chat, ipfs.team, protocol.ai
• TeamCity CI: http://ci.ipfs.team:8111

Getting started

We use a tool called Provsn to maintain the setup of hosts and services. The fundamental principle of Provsn is that hosts are in a certain state, and units of code are run to transition into a different state.

Provsn is a plain shell script, and each unit consists of shell scripts too:

• The env script exposes variables and functions to the unit itself, and other units.
• The build script is run on the client and builds container images, config files, etc.
• The install script is run on the host and transitions it into the desired state.

Note: there are a few bits of Ansible code left over, which are to be migrated to Provsn. You can find them in the ansible/ directory.

To test whether you're all set up, execute a simple command on all hosts.

> ./provsn exec all 'whoami'
pluto: root
uranus: root
[...]

Two environment variables can be used to alter Provsn's operation:

• PROVSN_JOBS -- this controls the number of hosts to run on in parallel, and defaults to 4.
• PROVSN_TRACE -- if set, this enables Bash tracing (set -x) for extensive debugging information. Note that this will contain sensitive information and secrets.

Usage

Known issues

• no verbose option, need to comment out dev-null-redirections in unit scripts
• if container that's supposed to be restarted is in a restart loop, we don't notice it's kinda running, and try to start it, and that fails because the name is already in use

Common tasks

• gathering ipfs debug info
• updating ipfs
• deploying a website
• adding a root user
• adding hashes to the blocklist

How can I get ssh access to the instances?

Add you ssh-key to the list of keys available in base/env.sh, like this: https://github.com/ipfs/infrastructure/blob/master/base/env.sh#L9 and then submit a PR with the changes.

Other community infrastructure:

More info in https://github.com/ipfs/community

• Github
  • https://github.com/ipfs
  • https://github.com/libp2p
  • https://github.com/ipld
  • https://github.com/multiformats
  • https://github.com/protocol
  • https://github.com/ipfsbot
• Communication
  • IRC: chat.freenode.net/#ipfs and https://chat.ipfs.io
  • ipfs-users group: https://groups.google.com/forum/#!forum/ipfs-users
  • Slack
  • https://twitter.com/ipfsbot
• CI / Testing
  • GitCop
  • TeamCity: http://ci.ipfs.team:8111
  • Travis CI
  • Circle CI

Contribute

Feel free to join in. All welcome. Open an issue!

This repository falls under the IPFS Code of Conduct.

Want to hack on IPFS?

License

MIT