Tangible Bytes

What I figured out (eventually)

1. I forgot that my Dockerfile contained a artisan route:cache line - this happened before my env was populated causing bad route caching
2. artisan route:list doesn’t show the cached routes - so it looks like it all ought to work
3. Env data is cached - better to use config in routes, and read the env vars in config/app.php than trying to read env vars in routes
4. Clearing caches didn’t fix anything

What fixed it

Set relevant env files

Add to config/app.php

    'root_domain' => env('ROOT_DOMAIN', 'test'),
    'admin_domain' => env('ADMIN_DOMAIN', 'know.test'),

in my routes files

// routes for my admin site like ..
Route::domain(config('app.admin_domain'))->get('/', [AuthenticatedSessionController::class, 'create'])
        ->name('login');


Route::domain('{site:domain}.' . config('app.root_domain'))->group(function () {
    // routes for public sites on a different subdomain here
});

NB each site has a domain entry in the sites table - which allows routing to lookup which site each subdomain is for/

Remove all caching steps from the Dockerfile

Instead I call a script

ENTRYPOINT ["/app/entrypoint.sh"]

#!/bin/bash


php artisan optimize

php artisan octane:frankenphp

This calls optimize at startup - when all env variables are present

Key Takeaway

Caching can be painful - artisan not showing the cached routes made it much harder.

Sometimes it’s tricky to debug containers - you can’t just change a file and restart because restarting is a new instance and you loose the change.

But with my optimize step in the right place it all worked OK

I still need to tweak my routes - and probably add the admin domain explicitly to all admin routes. Otherwise I may end up with conflicts - or public users accidentally requesting pages from the admin site.