Aller au contenu principal

Plan and setup the infrastructure for tests or production

Learn how to prepare your server infrastructure to deploy LinTO

Prerequistes​

Note

Our service stack relies on Docker Swarm Mode for orchestrating containers and services. This tutorial treats your infrastructure as a multi-host cluster. If you deploy the stack in a mono-host Docker Swarm Cluster (wich is obviously not recommanded for production) make sure to read the

Docker swarm mode​

  • 1 - The first machine of your cluster will become a docker swarm manager node. 
docker swarm init --advertise-addr <My private IP for cluster communication>
  • 2 - Get a join token for subsequent worker nodes
docker swarm join-token worker
  • 3 - This command outputs something like :
docker swarm join --token SWMTKN-1-xxx<Manager IP for cluster communication>:<Some TCP port used by docker swarm>

--> Use this command on every other machine that needs to join your docker swarm cluster

  • 4 - List nodes in the cluster hostname Private Ip for nodes communication
for NODE in $(docker node ls --format '{{.Hostname}}');
do echo -e "$(docker node inspect --format '{{.Status.Addr}}' "${NODE}") "${NODE}""; done
  • 5 - Put the corresponding pairs in /etc/hosts of every cluster node to enable domain name resolution between cluster's nodes.

Shared storage​

A good design pattern for highly available applications is to deploy the application as a container on a Docker Swarm cluster with persistent storage provided by GlusterFS. GlusterFS is a fast shared filesystem that can keep the container volume in sync between multiple nodes of the Docker Swarm cluster. In the event a node dies, Docker Swarm will spin up the container on another node. GlusterFS will ensure the container has access to the same data when it comes up.

In our case, Platform-Service-Manager will need read / write access on a shared folder made available on every node of your cluster for AI models used by transcription services. In this tutorial we will use a GlusterFS share folder.

Create directories for GlusterFS storage​

Setup the glusterFS directories where the gluster “bricks” (replicated volume) will reside. 

mkdir -p ~/linto_shared/data
mkdir -p ~/linto_shared/data
... and so on on every node of your swarm cluster

Note for single Machine Deployement: If you do not intent to add other machines to the cluster later, you can stop right here and jump to Server Installation.

Install Gluster FS​

Here, we use GLusterFS v7.0

Run this on every cluster's nodes

wget -O - https://download.gluster.org/pub/gluster/glusterfs/7/rsa.pub | sudo apt-key add -
DEBID=$(grep 'VERSION_ID=' /etc/os-release | cut -d '=' -f 2 | tr -d '"')
DEBVER=$(grep 'VERSION=' /etc/os-release | grep -Eo '[a-z]+')
DEBARCH=$(dpkg --print-architecture)
echo deb https://download.gluster.org/pub/gluster/glusterfs/LATEST/Debian/${DEBID}/${DEBARCH}/apt ${DEBVER} main | sudo tee --append /etc/apt/sources.list.d/gluster.list
sudo apt-get install && sudo apt-get install glusterfs-server -y
sudo systemctl enable --now glusterd

Peer Gluster FS nodes​

On the swarm cluster manager node, supposedly the first "brick" of the distributed and replicated file storage, run :

sudo gluster peer probe <HostName-X for node X of the cluster>

--> Use this command for every other cluster's nodes hostnames.

Setup the replicated volume​

sudo gluster volume create linto_shared \
replica <Number of cluster nodes> \
<First Node HostName>:<corresponding folder> \
<Second NodeHostName>:<corresponding folder> \
... (etc)
force

--> Use "force" directive if you want to use some folder inside the node root folder.

Start the replicated volume​

sudo gluster volume start linto_shared

Prepare mount points for the volume on each node​

mkdir ~/linto_shared_mount

This is where the services will access the filesystem provided by the Gluster FS volume

Prepare automount shared volume on each node (effective after reboot)​

Use this wisely and don't break your /etc/fstab

command will look like :

echo 'localhost:/linto_shared <Local mount point> glusterfs defaults,_netdev,backupvolfile-server=localhost 0 0' | sudo tee --append /etc/fstab
mount -a

Mount your volume on each node​

sudo mount.glusterfs localhost:/shared_linto <Local mount point>