deuxfleurs.fr

Many things are still missing here, including a proper documentation. Please stay nice, it is a volunter project. Feel free to open pull/merge requests to improve it. Thanks.

Our abstraction stack

We try to build a generic abstraction stack between our different resources (CPU, RAM, disk, etc.) and our services (Chat, Storage, etc.), we develop our own tools when needed:

• garage: S3-compatible lightweight object store for self-hosted geo-distributed deployments (we also have a legacy glusterfs cluster)
• diplonat: network automation (firewalling, upnp igd)
• bottin: authentication and authorization (LDAP protocol, consul backend)
• guichet: a dashboard for our users and administrators
• ansible: physical node configuration
• nomad: schedule containers and handle their lifecycle
• consul: distributed key value store + lock + service discovery
• stolon + postgresql: distributed relational database
• docker: package, distribute and isolate applications

Some services we provide:

• Websites: garage (static) + fediverse blog (plume)
• Chat: Synapse + Element Web (Matrix protocol)
• Email: Postfix SMTP + Dovecot IMAP + opendkim DKIM + Sogo webmail (legacy) | Alps webmail (experimental)
• Storage: Seafile (legacy) | Nextcloud (experimental)
• Visio: Jitsi

As a generic abstraction is provided, deploying new services should be easy.

I am lost, how this repo works?

To ease the development, we make the choice of a fully integrated environment

1. os the base os for the cluster
   1. build: where you will build our OS image based on Debian that you will install on your server
   2. config: our Ansible recipes to configure and update your freshly installed server
2. apps apps we deploy on the cluster
   1. build: our Docker files to build immutable images of our applications
   2. integration: Our Docker compose files to test locally how our built images interact together
   3. config: Files containing application configurations to be deployed on Consul Key Value Store
   4. deployment: Files containing application definitions to be deployed on Nomad Scheduler
3. op_guide: Guides to explain you operations you can do cluster wide (like configuring postgres)

Start hacking

Deploying/Updating new services is done from your machine

The following instructions are provided for ops that already have access to the servers (meaning: their SSH public key is known by the cluster).

Deploy Nomad on your machine:

export NOMAD_VER=1.0.1
wget https://releases.hashicorp.com/nomad/${NOMAD_VER}/nomad_${NOMAD_VER}_linux_amd64.zip
unzip nomad_${NOMAD_VER}_linux_amd64.zip
sudo mv nomad /usr/local/bin
rm nomad_${NOMAD_VER}_linux_amd64.zip

Deploy Consul on your machine:

export CONSUL_VER=1.9.0
wget https://releases.hashicorp.com/consul/${CONSUL_VER}/consul_${CONSUL_VER}_linux_amd64.zip
unzip consul_${CONSUL_VER}_linux_amd64.zip
sudo mv consul /usr/local/bin
rm consul_${CONSUL_VER}_linux_amd64.zip

Create an alias (and put it in your .bashrc) to bind APIs on your machine:

alias bind_df="ssh \
  -p110 \
  -N \
  -L 1389:bottin2.service.2.cluster.deuxfleurs.fr:389 \
  -L 4646:127.0.0.1:4646 \
  -L 5432:psql-proxy.service.2.cluster.deuxfleurs.fr:5432 \
  -L 8082:traefik-admin.service.2.cluster.deuxfleurs.fr:8082 \
  -L 8500:127.0.0.1:8500 \
  <a server from the cluster>"

and run:

bind_df

Adrien uses .ssh/config configuration instead. I works basically the same. Here it goes:

# in ~/.ssh/config 

Host deuxfleurs
    User adrien
    Hostname deuxfleurs.fr
    # If you don't use the default ~/.ssh/id_rsa to connect to Deuxfleurs
    IdentityFile <some_key_path>
    PubKeyAuthentication yes
    ForwardAgent No
    LocalForward 1389 bottin2.service.2.cluster.deuxfleurs.fr:389
    LocalForward 4646 127.0.0.1:4646
    LocalForward 5432 psql-proxy.service.2.cluster.deuxfleurs.fr:5432
    LocalForward 8082 traefik-admin.service.2.cluster.deuxfleurs.fr:8082
    LocalForward 8500 127.0.0.1:8500

Now, to connect, do the following:

ssh deuxfleurs -N