Deploy to Supabase.com
This guide explains the current process for deploying and maintaining workers in hosted environment.
Deploying workers
Workers are just regular Edge Functions that run for few minutes and auto-respawn when CPU or Wall clock limits are reached (read more on those Limits | Supabase Docs).
Deploying new version is straightforward - you just update worker code
and call npx supabase functions deploy worker-name to deploy it.
When deploying a new version of your worker, be aware of the following:
Starting Workers
Just like in local development, workers are started with a simple HTTP request. However, you need to pass Authorization header with your ANON_KEY when calling function. Consult this Supabase Docs page for details: Invoking remote functions
After the initial request, the worker will auto-respawn if terminated.
Stopping Workers
Currently, there’s no built-in way to gracefully stop a running worker. You need to remove it from the Supabase Dashboard and wait for its CPU or clock limits to be reached for it to stop.
This will change in future releases.
Ensuring Worker Availability
To ensure your worker stays active, it’s recommended to use Cron to ping your worker edge function every few seconds. This is recommended in case there are any bugs in the respawn process, which currently is triggered from withing the worker that is shutting down.
Use this Supabase guide Invoking Supabase Function ever few seconds.
but use 5 seconds instead of 30 seconds.