On top of providing security via query complexity analysis and query depth analysis.
I believe having query timeouts is an important feature.

Feature Request Type

• Core functionality
• Alteration (enhancement/optimization) of existing feature(s)
• New behavior

Description

Backend design is a hard process and inevitably involves introducing unexpected and hard to debug issues
especially involving the database or filesystem that it is pulling information from.

Amatuer devs could write SQL queries that may halt the DB or perform file operations that are not sustainable.

So I propose having the following feature set:

1. Having the query timeout and kill the respective event_loop would be a safer option in these cases.
2. In addition, there could be a propagate timeout function that would run a cleanup of an external DB query that's taking too long to execute.
3. Since returning a GraphQL error may break the spec. We could return a plain http reponse with a custom 5xx error.
4. Global default timeout
5. Granular timeout control, meaning, we can define a safe timeout for each query.

I'd love to know what the rest of the community thinks about this suggestion.