Add more info on how to set up linkup (#21)

README.md

@@ -56,7 +56,7 @@ services:
   - name: backend
     remote: https://api-dev.hosting-provider.com
     local: http://localhost:9000
-    directory: ./
+    directory: ./relative/path/to/backend
 domains:
   - domain: dev-domain.com
     default_service: web
@@ -67,9 +67,22 @@ domains:
     default_service: backend
 ```
 
+### Local environment variables
+
+When booting local servers to run in linkup, they must be configured with environment variables that point to your linkup domain.
+
+To do this, linkup appends the contents of `.env.linkup` to the `.env` file located in the `directory` configuration field of the service.
+
+Linkup will fail to start if there is no `.env.linkup` file, and it will warn you to restart your local server if it was already booted on `linkup start`.
+
 ## Deploying Linkup
 
-In order to run linkup sessions, you need a dedicated domain for linkup to run on and deployed copies of the remote services you want to provide.
+In order to run linkup sessions, you need:
+
+- A dedicated domain for the linkup cloudflare worker to run on
+- Deployed copies of the remote services you want to provide
+
+### Configuring the domain & worker
 
 Linkup is deployed as a cloudflare worker with a key-value store, and can be deployed using the wrangler cli:
 
@@ -79,3 +92,13 @@ cp wrangler.toml.sample wrangler.toml
 # Edit wrangler.toml to point to your cf kv store
 npx wrangler@latest deploy
 ```
+
+It is also easiest to use a domain in cloudflare. Set the `*` and `*.*` subdomains to point to the worker you just deployed.
+
+### Deploying remote services
+
+The remote services you would like to make available to linkup sessions have a few requirements:
+
+- They must be accessible at a public url
+- Their environment variables must be configured to point to your linkup domain
+- For requests _from_ the remote service to be correctly routed to linkup sessions, they must be able to [propagate trace contexts](https://www.w3.org/TR/trace-context/) through requests made from the service. The easiest way to acheive this is to use an OpenTelemetry client library to instrument your http client. Here is [an example for javascript](https://www.npmjs.com/package/@opentelemetry/instrumentation-http).