hasura/heal-remote-schema repository overview

⁠Heal Remote Schemas

A script for auto healing the inconsistent remote schemas in graphql-engine.

⁠Configurable parameters:

This script uses the following configurable parameters (environment variables) for running the auto-healing node server:

1. GRAPHQL_ENGINE_URL: The url of the graphql engine (e.g. http://localhost:8080)
2. HASURA_GRAPHQL_ADMIN_SECRET: The admin secret of the graphql engine.
3. CHECK_INTERVAL_SECONDS: The interval for checking the metadata for any inconsistencies (in seconds).
4. RELOAD_INTERVAL_SECONDS: The interval for reloading the remote schemas (in seconds).

⁠How does this work?

The checking thread runs the following query to the graphql-engine every CHECK_INTERVAL_SECONDS:

POST v1/metadata
admin

{
  "type": "get_inconsistent_metadata",
  "args": {}
}

If the above query returns any inconsistent remote schema, it stores them in an array.

Now, the reloading thread checks if the inconsistent remote schema array is empty every RELOAD_INTERVAL_SECONDS. If it is empty, then it will just log:

All remote schemas are consistent.

Now, if the array is not empty, then it will try reloading the inconsistent remote schemas:

POST v1/metadata
admin

{
  "type" : "reload_metadata",
  "args": {
    "reload_remote_schemas": *inconsistentSchemas
  }
}

To run this server locally:

⁠Via node

npm install
touch .env
npm start

⁠Via docker

The image is also pushed to docker hub. So you can directly use that in your docker-compose.yaml file or run it separately via docker run.

version: '3.6'
services:
  heal-rs:
    image: paritosh08/heal-remote-schema
    environment:
      NODE_ENV: production
      GRAPHQL_ENGINE_URL: http://graphql-engine:8080
      HASURA_GRAPHQL_ADMIN_SECRET: hasura
      CHECK_INTERVAL_SECONDS: 10
      RELOAD_INTERVAL_SECONDS: 1
  graphql-engine: ...