S3 content indexing becomes slow at higher load

[dzikoysk]

What happened?

Browsing repository content in the UI causes a lot of requests to the S3 service. It has a few drawbacks:

• Extra requests to the S3 service increase general response time
• S3 service may rate limit requests/timeout, so file browser stops working
• Handling so many in highly concurrent environment may cost us more than a simple cache
• Cloud providers usually charge users for the number of requests, so it'd be nice if we could slightly limit them

Some sort of cache layer would probably solve the issue, but we have to be careful, because:

1. It may increase resources usage of Reposilite, especially for the smallest private repositories (minimal requirements)
2. What's even more important - the cache state may become outdated, especially between 2+ Reposilite's instances (k8s)

For now, I think it makes sense to hide this behind a feature flag - caching should be opt-in, with configurable TTL

Reposilite version

3.x

Relevant log output

No response