Is it possible to cache based on request body?

[KaiserKarel]

Ideally I'd like to first receive the request body and use that to generate the CacheKey. Is this currently possible without forking Pingora? I wonder if it fits in the current lifecycle management.

[Object905]

Interested in that too. For transparent caching of graphql requests.
Latency would go up due to buffering for sure.

[KaiserKarel]

Interested in that too. For transparent caching of graphql requests. Latency would go up due to buffering for sure.

Similar situation for me, except caching JSON-RPC.

[KaiserKarel]

@johnhurt @gumpt how would we like this feature to be added? Reuse the ProxyHttp trait, and add another service which has a different lifecycle?

Another usecase we have is that some upstreams support additional methods (JSON-RPC), so we'd like to determine the upstream based on the request body too.