Skip to content

Latest commit

 

History

History
61 lines (38 loc) · 2.5 KB

create-an-instance.md

File metadata and controls

61 lines (38 loc) · 2.5 KB
Raw

Create a LaminDB instance on LaminHub

⚠️ Currently, these instructions are for customers only.

This guide explains how to create a LaminDB instance that's fully deployed in LaminHub and comes with default SQL permission roles.

Two things happen in addition to the open-source lamin init:

  • connecting the LaminHub backend to your Postgres database
  • creating SQL permission roles read, write, and admin and synching them with storage permissions

Prerequisites

Database server

Ensure you have access to a PostgreSQL server with a default database called postgres and a default superuser called postgres.

Ensure that our LaminHub backend servers are whitelisted so that our servers can access your database. The IP addresses are listed here) resoled by AWS region.

Storage

You can optionally provide your S3 bucket to serve as the default storage location of a LaminDB instance.

To give Lamin access to your bucket, attach this AWS permission policy (here is how to do this manually through the AWS console).

Endpoint Parameters

  • API_URL: The URL of the API, depending on your region, pick from here
  • DB_SERVER_URL: The URL for the PostgreSQL server in the format postgresql://postgres:<password>@<host>:<port>/postgres
  • NAME: The name of the LaminDB instance

Optionals:

  • SCHEMA: The list of schemas you want to mount, e.g., bionty,wetlab
  • STORAGE: The S3 bucket URI in the format s3://my-bucket

Create the instance

Install lamindb_setup via pip install lamindb_setup in case you don't have a lamindb installation.

In your terminal, run lamin login to ensure you're authenticated.

To create the instance, run the following python code:

import requests
from lamindb_setup import settings

API_URL = "https://us-west-2.api.lamin.ai"  # Update API_URL according to the region where you to deploy.
DB_SERVER_URL = "postgresql://postgres:<PASSWORD>@<HOST>:<PORT>/postgres"
NAME = "YOUR-INSTANCE-NAME"
SCHEMA_STR = "bionty"
STORAGE = "s3://YOUR-BUCKET-NAME"

response = requests.put(
    f"{API_URL}/instances?db_server_url={DB_SERVER_URL}&name={NAME}&schema_str={SCHEMA_STR}&storage={STORAGE}",
    headers={"authentication": f"Bearer {settings.user.access_token}"},
)

Upon success, navigate to https://lamin.ai/handle/instance-name, where handle is your user handle and instance-name the NAME parameter above.