I'm developing a fullstack application using FastAPI for the backend. In this application, one or more users can query (read) a file, while multiple administrators can modify (write) it simultaneously. I'm concerned about the possibility of race conditions and other concurrency issues that might arise due to these simultaneous operations, but I don't know how to implement it in an async way, I have only used threading and multiprocessing.
I would appreciate any advice, best practices, and an implementation example to address these concurrency issues in my fullstack application. Thank you!
FastAPI is a bit out of scope for this specific problem, and doesn't have a way or mechanism specifically designed for this matter. You would need to decide what is the best way to mitigate race conditions for files generally, regardless if it's FastAPI in Python, or any other technology or language.
Do you want to lock a file for write operations (using a read-write lock) or do you want users to edit the same file at the same time (which can get pretty complicated)? If only one write is allowed, then it's just a read-write lock (i.e. multiple reads allowed, but writes are exclusive) involved, possibly with versioning, to prevent users writing outdated files. This can be implemented using a database.
It's important to use a database or a persistent storage if you will scale up to more than one instance of your backend app. In this case, the 2 or more instances need to share the same locks globally. Otherwise threading and basic locks in Python should suffice.
Probably more details of how the files should be used, especially for writes, would allow us to provide better and more detailed solutions.