Skip to content

Latest commit

 

History

History
128 lines (100 loc) · 2.91 KB

README.md

File metadata and controls

128 lines (100 loc) · 2.91 KB
Raw

Server-side caching plugin

A simple plugin for Hasura DDN server-side request caching.

Installation

Starting the plugin

The plugin can be run for local development using docker:

$ docker build -t caching-plugin .
$ docker run -p 8787:8787 caching-plugin

Adding the plugin to a Hasura project

Add the following configuration to one of your Hasura DDN subgraphs. The first block describes the "pre-parse" stage, which will return the cached response if it exists and the given query is marked for caching.

kind: LifecyclePluginHook
version: v1
definition:
  pre: parse
  name: cache_get_test
  url:
    value: http://localhost:8787
  config:
    request:
      headers:
        additional:
          hasura-m-auth:
            value: my-secret-header
          hasura-plugin-phase:
            value: pre-parse
      rawRequest:
        query: {}
        variables: {}

The second block describes the "pre-response" stage, which will cache the response if it hasn't already been cached and if the given query is marked for caching.

kind: LifecyclePluginHook
version: v1
definition:
  pre: response
  name: cache_set_test
  url:
    value: http://localhost:8787
  config:
    request:
      headers:
        additional:
          hasura-m-auth:
            value: my-secret-header
          hasura-plugin-phase:
            value: pre-response
      rawRequest:
        query: {}
        variables: {}

Configuration

Configuration is found in src/config.js. Currently, there are three available keys:

Expected headers

headers: { "hasura-m-auth": "my-secret-header" }

Headers expected from the engine when a request to the plugin is made. Currently, only one header is expected: hasura-m-auth is a user-specified secret that the engine should send over in its requests.

A list of queries_to_cache

queries_to_cache: [
  "query { Artist { Name } }",
  "query { Album { Artist { Name } } }",
];

When a user makes a request to the server, the caching plugin will parse the request into an AST, and then check to see whether the AST matches the AST of any of these queries. If it does, the query will be cached. This means that whitespace is unimportant in these queries, and you can write them over multiple lines.

A result will be cached for each set of session and query variables that use this same query template.

The time_to_live for a cache entry

time_to_live: 600;

The length of time after which a given cache entry should be invalidated. This is measured in seconds.

A redis_url for storing the cache entries.

redis_url: "redis://redis:6379";

An otel_endpoint for OpenTelemetry traces.

otel_endpoint: "http://jaeger:4318/v1/traces";

Any other otel_headers for the trace collectotr.

otel_headers: {
  Authorization: "pat <my-personal-access-token>";
}