Rate limiting middleware framework for actix-web
This crate provides an asynchronous and concurrent rate limiting middleware based on actor model which can be wraped around an Actix application. Middleware contains a store which is used to identify client request.
Check out the documentation here.
Comments, suggesstions and critiques are welcome!
Add this to your Cargo.toml:
[dependencies]
actix-ratelimit = "0.3.1"
Version 0.3.*
supports actix-web v3. If you're using actix-web v2, consider using version 0.2.*
.
Minimal example:
use actix_web::{web, App, HttpRequest, HttpServer, Responder};
use actix_ratelimit::{RateLimiter, MemoryStore, MemoryStoreActor};
use std::time::Duration;
async fn greet(req: HttpRequest) -> impl Responder{
let name = req.match_info().get("name").unwrap_or("World!");
format!("Hello {}!", &name)
}
#[actix_web::main]
async fn main() -> std::io::Result<()> {
// Initialize store
let store = MemoryStore::new();
HttpServer::new(move ||{
App::new()
// Register the middleware
// which allows for a maximum of
// 100 requests per minute per client
// based on IP address
.wrap(
RateLimiter::new(
MemoryStoreActor::from(store.clone()).start())
.with_interval(Duration::from_secs(60))
.with_max_requests(100)
)
.route("/", web::get().to(greet))
.route("/{name}", web::get().to(greet))
})
.bind("127.0.0.1:8000")?
.run()
.await
}
Sending a request returns a response with the ratelimiting headers:
$ curl -i "http://localhost:8000/"
HTTP/1.1 200 OK
content-length: 13
content-type: text/plain; charset=utf-8
x-ratelimit-remaining: 99
x-ratelimit-reset: 52
x-ratelimit-limit: 100
date: Tue, 04 Feb 2020 21:53:27 GMT
Hello World!
Exceeding the limit returns HTTP 429 Error code.
A store is a data structure, database connection or anything which can be used to store ratelimit data associated with a client. A store actor which acts on this store is responsible for performiing all sorts of operations(SET, GET, DEL, etc). It is Important to note that there are multiple store actors acting on a single store.
memory
(in-memory store based on concurrent hashmap)redis-store
(based on redis-rs)memcached
(based on r2d2-memcache, see note to developers below)
To implement your own store, you have to implement an Actor which can handle ActorMessage type and return ActorResponse type. Check the module level documentation for more details and a basic example.
- By default, all features are enabled. To use a particular feature, for instance redis, put this in your Cargo.toml:
[dependencies]
actix-ratelimit = {version = "0.3.1", default-features = false, features = ["redis-store"]}
- By default, the client's IP address is used as the identifier which can be customized using ServiceRequest instance. For example, using api key header to identify client:
#[actix_web::main]
async fn main() -> std::io::Result<()> {
// Initialize store
let store = MemoryStore::new();
HttpServer::new(move ||{
App::new()
.wrap(
RateLimiter::new(
MemoryStoreActor::from(store.clone()).start())
.with_interval(Duration::from_secs(60))
.with_max_requests(100)
.with_identifier(|req| {
let key = req.headers().get("x-api-key").unwrap();
let key = key.to_str().unwrap();
Ok(key.to_string())
})
)
.route("/", web::get().to(greet))
.route("/{name}", web::get().to(greet))
})
.bind("127.0.0.1:8000")?
.run()
.await
}
-
The memcache store uses a separate key to keep track of expiry since there's no way to get ttl of keys in memcache natively yet. This means memcache store will use double the number of keys as compared to redis store. If there's any better way to do this, please considering opening an issue!
-
It is important to initialize store before creating HttpServer instance, or else a store will be created for each web worker. This may lead to instability and inconsistency! For example, initializing your app in the following manner would create more than one stores:
#[actix_web::main]
async fn main() -> std::io::Result<()> {
HttpServer::new(move ||{
App::new()
.wrap(
RateLimiter::new(
MemoryStoreActor::from(MemoryStore::new()).start())
.with_interval(Duration::from_secs(60))
.with_max_requests(100)
)
.route("/", web::get().to(greet))
.route("/{name}", web::get().to(greet))
})
.bind("127.0.0.1:8000")?
.run()
.await
}
- To enable ratelimiting across multiple instances of your web application(multiple http servers behind load balancer), consider using a feature called
session stickiness
supported by popular cloud services such as AWS, Azure, etc.
This project has not reached v1.0, so some instability and breaking changes are to be expected till then.
You can use the issue tracker in case you encounter any problems.
This project is licensed under MIT license.
License: MIT