Skip to content

Hosting locally for development

For development purposes, Prisme.ai can be started in two ways:

  1. With docker only, including databases images (Elasticsearch, Redis, MongoDB)
  2. With a mix of Docker & NodeJS for a developer localhost environment

Both ways are meant for testing only and not for a production setup.

Docker

Requirements

Dependencies :

  • git
  • Docker 20+
  • docker-compose

Hardware :

  • Minimum disk space for docker images : 20GB+
  • Minimum disk space for volumes : 5GB+, ideally 50GB
  • Recommended specs : 4 vCPU / 16GB RAM
  • Wildcard DNS pointing to the Docker host machine
    • Needed by pages URL which are built with the workspace slug as a subdomain

Configuration

  1. git clone https://gitlab.com/prisme.ai/prisme.ai.git
    By default, the docker-compose.yml & .env files are configured to run on a local setup, served through 3 localhost DNS which we define in our /etc/hosts file.
    In order for the platform to be available from whatever online network, these domains would have to be reconfigured as described below.
    Databases-wise, this default local setup is also configured to use databases from the docker-compose.yml, using Docker network links and persisting data in the data/ root directory.
    All available environment variables are described here.

  2. Open & edit the root .env file, configuring a few shared environment variables used by docker-compose.yml :

PAGES_HOST : Pages wildcard domain, prefixed with a beginning dot.
If PAGES_HOST = .pages.local.prisme.ai:3100, a page dashboard within a workspace collections will be available at http[s]://collections.pages.local.prisme.ai:3100/dashboard.
Here, both collections.pages.local.prisme.ai and anyOtherWorkspace.pages.local.prisme.ai must point to the host docker machine IP through a wildcard DNS record like this one :

*.pages.local.prisme.ai 10800 IN CNAME studio.local.prisme.ai.
# or ...
*.pages.local.prisme.ai 10800 IN A 100.100.100.100
Where either studio.local.prisme.ai DNS or 100.100.100.100 IP point to the host docker machine.

CONSOLE_URL : Public URL serving the studio itself. Used by the 4 following services : api-gateway, console, pages, runtime.
On a single machine setup with Docker, this is just another domain pointing to the same IP as PAGES_HOST does, which could even be the same parent domain as PAGES_HOST since both services (Pages & Studio) will be served through different ports anyway. Used by the 4 following services : api-gateway, console, pages, runtime.

API_URL : Public URL serving the API, ended with /v2.
Again, on a single machine setup with Docker, this is just another domain pointing to the same IP as PAGES_HOST and CONSOLE_URL do, served through a third port.

Other environment variables which might be tuned from the start :

  • ACCOUNT_VALIDATION_METHOD : defaults to 'email', set it to 'auto' or 'manual' if you don't have any mailgun api key available (api-gateway)
  • UPLOADS_MAX_SIZE : defaults to '10mb', can be set to '100mb' or bigger to allow processing bigger files (api-gateway, workspaces, runtime)

See full description of available environments variable here.

Start

docker-compose up

Studio will be available at the defined CONSOLE_URL, by default http://studio.local.prisme.ai:3000

NodeJS

Dependencies :

  • NodeJS 20+
  • npm 10+
  • git
  • Docker 20+
  • docker-compose

Hardware :

  • Minimum disk space for docker images : 15GB+
  • Minimum disk space for volumes : 5GB+, ideally 50GB
  • Recommended specs : 4 vCPU / 16GB RAM

Start

If willing to change any part of this codebase, we recommend testing your updates live using our npm start script, which starts the services you want to change with NodeJS, and everything else with Docker :

git clone https://gitlab.com/prisme.ai/prisme.ai.git  
npm install  
npm start  # Unselect services with spacebar & arrow keys in order to start them from NodeJS 

The npm start will prompt whether you want to start services from build images or local repository using NodeJS.

In order to run Prisme.ai from its official Docker images, make sure that all services have their box ticked & type 'Enter' : this will start each service (including databases) from their own docker-compose.yml (located in each services/* directory).

If willing to develop or simply start using NodeJS, untick all services' boxes by pressing 'a' and 'Enter' : this will start databases using Docker, and individual services with their respective npm run dev command.

Whether you started using Docker or NodeJS, all persistent data will be stored inside the data/ directory at the root of your repository.

By default, studio will be available at http://studio.local.prisme.ai:3000