How do I avoid body_limit error in boost's beast and correctly handle large messages

问题

I have cases when there is a chunked response which is too big for beast, and I want to stop before I get to beast's body_limit, and continue handling the message from that point using plain boost::asio. Mind that this (obviously) means I already received the header and a large part of the body.

I'm using it for a reverse proxy, so basically what I want to do is somehow send the incomplete response to the http client, while continuing relaying the remaining response data using boost::asio.

I'm guessing I'll need to somehow serialize the incomplete response, maybe using operator<< to std::stringstream, send that to the client using boost::asio, and continue the communication from there.

Will this work? Is this the correct way of doing that, or is there a better way, maybe even using beast api? Is there another way to handle chunked messages that are about to exceed body_limit in beast's api?

Thanks in advance, David.

UPDATE

I finally abandoned the idea of falling back to boost asio, and am now trying to receive the http message (chunked or regular) in chunks with a fixed size buffer, so that I don't reach body limit. I'm just done skimming over Receive/parse the message body one chunk at a time · Issue #154 · boostorg/beast, and it seems that it's exactly what I need. I'm trying to implement a reverse proxy as well.. I tried to use Incremental Read 💡 - 1.70.0 but get a Reference to non-static member function must be called error when trying to compile this line:

ctx->response.get().body().data = response_buffer;

Maybe the incremental read example page is not updated with the latest syntax? Do you have an example relevant for the reverse proxy I'm trying to write?

Thanks in advance, David

回答1:

Maybe the incremental read example page is not updated with the latest syntax? Do you have an example relevant for the reverse proxy I'm trying to write?

The examples in the docs are compiled, so they can't possibly be out of date. Perhaps you are mixing different versions of the example and Beast? Are you using http::buffer_body? What does the declaration of your message look like?

回答2:

By default, Beast's parser limits the size of the body to 1MB for requests and 8MB for responses. This is to prevent trivial resource exhaustion attacks. You can always increase the limit, or eliminate it entirely (by setting it to the largest uint64_t) by calling parser::body_limit : https://www.boost.org/doc/libs/1_71_0/libs/beast/doc/html/beast/ref/boost__beast__http__parser/body_limit.html

来源：https://stackoverflow.com/questions/58556690/how-do-i-avoid-body-limit-error-in-boosts-beast-and-correctly-handle-large-mess