Falcon is a multi-process, multi-fiber rack-compatible HTTP server built on top of async, async-io, async-container and async-http. Each request is executed within a lightweight fiber and can block on up-stream requests without stalling the entire server process. Falcon supports HTTP/1 and HTTP/2 natively.
Priority Business Support is available.
Initially, when I developed async, I saw an opportunity to implement async-http: providing both client and server components. After experimenting with these ideas, I decided to build an actual web server for comparing and validating performance primarily out of interest. Falcon grew out of those experiments and permitted the ability to test existing real-world code on top of async.
Once I had something working, I saw an opportunity to simplify my development, testing and production environments, replacing production (Nginx+Passenger) and development (Puma) with Falcon. Not only does this simplify deployment, it helps minimize environment-specific bugs.
My long term vision for Falcon is to make a web application platform which trivializes server deployment. Ideally, a web application can fully describe all it's components: HTTP servers, databases, periodic jobs, background jobs, remote management, etc. Currently, it is not uncommon for all these facets to be handled independently in platform specific ways. This can make it difficult to set up new instances as well as make changes to underlying infrastructure. I hope Falcon can address some of these issues in a platform agnostic way.
As web development is something I'm passionate about, having a server like Falcon is empowering.
Add this line to your application's Gemfile:
gem 'falcon'
And then execute:
$ bundle
Alternatively, install in terminal:
$ gem install falcon
You can run falcon serve
directly. It will load the config.ru
and start serving on https://localhost:9292. Please try the interactive online tutorial.
The falcon serve
command has the following options for you to use:
$ falcon --help
falcon [--verbose | --quiet] [-h/--help] [-v/--version] <command>
An asynchronous HTTP server.
[--verbose | --quiet] Verbosity of output for debugging.
[-h/--help] Print out help information.
[-v/--version] Print out the application version.
<command> One of: serve, virtual. Default: serve
serve [-b/--bind <address>] [-p/--port <number>] [-h/--hostname <hostname>] [-t/--timeout <duration>] [--reuse-port] [-c/--config <path>] [--forked | --threaded | --hybrid] [-n/--count <count>] [--forks <count>] [--threads <count>]
Run an HTTP server.
[-b/--bind <address>] Bind to the given hostname/address Default: https://localhost:9292
[-p/--port <number>] Override the specified port
[-h/--hostname <hostname>] Specify the hostname which would be used for certificates, etc.
[-t/--timeout <duration>] Specify the maximum time to wait for blocking operations. Default: 60
[--reuse-port] Enable SO_REUSEPORT if possible. Default: false
[-c/--config <path>] Rackup configuration file to load Default: config.ru
[--forked | --threaded | --hybrid] Select a specific parallelism model Default: forked
[-n/--count <count>] Number of instances to start. Default: 8
[--forks <count>] Number of forks (hybrid only).
[--threads <count>] Number of threads (hybrid only).
To run on a different port:
$ falcon serve --port 3000
Falcon can replace Nginx as a virtual server for Ruby applications. This is an experimental feature.
/--------------------\
| Client Browser |
\--------------------/
||
(TLS + HTTP/2 TCP)
||
/--------------------\
| Falcon Proxy (SNI) |
\--------------------/
||
(HTTP/2 UNIX PIPE)
||
/--------------------\
| Application Server | (Rack Compatible)
\--------------------/
You need to create a falcon.rb
configuration in the root of your application, and start the virtual host:
$ cat /srv/http/example.com/falcon.rb
#!/usr/bin/env -S falcon host
load :rack, :lets_encrypt_tls, :supervisor
rack 'hello.localhost', :lets_encrypt_tls
supervisor
% falcon virtual /srv/http/example.com/falcon.rb
The falcon virtual server is hard coded to redirect http traffic to https, and will serve each application using an internal SNI-based proxy.
Falcon works perfectly with rails
apps.
-
Add
gem 'falcon'
to yourGemfile
and perhaps removegem 'puma'
once you are satisfied with the change. -
Run
falcon serve
to start a local development server.
Alternatively run RACK_HANDLER=falcon rails server
to start the server (at least, until rack#1181 is merged).
With older versions of Rails, the Rack::Lock
middleware can be inserted into your app by Rails. Rack::Lock
will cause both poor performance and deadlocks due to the highly concurrent nature of falcon
. Other web frameworks are generally unaffected.
Please ensure you specify config.threadsafe!
in your config/application.rb
:
module MySite
class Application < Rails::Application
# ...
# Enable threaded mode
config.threadsafe!
end
end
Please ensure you have config.allow_concurrency = true
in your configuration.
This became the default in Rails 5 so no change is necessary unless you explicitly disabled concurrency, in which case you should remove that configuration.
Falcon supports (partial and full) rack.hijack
for both for HTTP/1 and HTTP/2 connections. You can thus use async-websocket in any controller layer to serve WebSocket connections.
The rack.hijack
functionality is compatible with ActionCable. If you use the async
adapter, you should run falcon in threaded mode, or in forked mode with --count 1
. Otherwise, your messaging system will be distributed over several processes with no IPC mechanism. You might like to try out async-redis as an asynchronous message bus.
Falcon supports the rack.early_hints
API when running over HTTP/2.
Falcon can restart very quickly and is ideal for use with guard. See guard-falcon for more details.
Falcon can run in the same process on a different thread, so it's great for use with Capybara (and shared ActiveRecord transactions). See falcon-capybara for more details.
You can invoke Falcon via rackup
:
rackup --server falcon
This will run a single-threaded instance of Falcon using http/1
. While it works fine, it's not recommended to use rackup
with falcon
, because performance will be limited.
In order to use curl
with self-signed localhost certificates, you need to specify --insecure
or the path of the certificate to validate the request:
> curl -v https://localhost:9292 --cacert ~/.localhost/localhost.crt
* Trying ::1...
* TCP_NODELAY set
* Connected to localhost (::1) port 9292 (#0)
* ALPN, offering http/1.1
* Cipher selection: ALL:!EXPORT:!EXPORT40:!EXPORT56:!aNULL:!LOW:!RC4:@STRENGTH
* successfully set certificate verify locations:
* CAfile: /Users/samuel/.localhost/localhost.crt
CApath: none
* TLSv1.2 (OUT), TLS header, Certificate Status (22):
* TLSv1.2 (OUT), TLS handshake, Client hello (1):
* TLSv1.2 (IN), TLS handshake, Server hello (2):
* TLSv1.2 (IN), TLS handshake, Certificate (11):
* TLSv1.2 (IN), TLS handshake, Server key exchange (12):
* TLSv1.2 (IN), TLS handshake, Request CERT (13):
* TLSv1.2 (IN), TLS handshake, Server finished (14):
* TLSv1.2 (OUT), TLS handshake, Certificate (11):
* TLSv1.2 (OUT), TLS handshake, Client key exchange (16):
* TLSv1.2 (OUT), TLS change cipher, Change cipher spec (1):
* TLSv1.2 (OUT), TLS handshake, Finished (20):
* TLSv1.2 (IN), TLS change cipher, Change cipher spec (1):
* TLSv1.2 (IN), TLS handshake, Finished (20):
* SSL connection using TLSv1.2 / ECDHE-RSA-AES256-GCM-SHA384
* ALPN, server accepted to use http/1.1
* Server certificate:
* subject: O=Development/CN=localhost
* start date: Aug 10 00:31:43 2018 GMT
* expire date: Aug 7 00:31:43 2028 GMT
* subjectAltName: host "localhost" matched cert's "localhost"
* issuer: O=Development/CN=localhost
* SSL certificate verify ok.
> GET / HTTP/1.1
> Host: localhost:9292
> User-Agent: curl/7.63.0
> Accept: */*
>
< HTTP/1.1 301
< location: /index
< cache-control: max-age=86400
< content-length: 0
<
* Connection #0 to host localhost left intact
Falcon uses an asynchronous event-driven reactor to provide non-blocking IO. It can handle an arbitrary number of in-flight requests with minimal overhead per request.
It uses one Fiber per request, which yields in the presence of blocking IO.
- Improving Ruby Concurrency – Comparison of Falcon and Puma.
Falcon uses a pre-fork model which loads the entire rack application before forking. This reduces per-process memory usage.
async-http has been designed carefully to minimize IO related garbage. This avoids large per-request memory allocations or disk usage, provided that you use streaming IO.
Falcon can be an important part of your business or project, both improving performance and saving money. As such, priority business support is available to make every project a success. The agreement will give you:
- Better software by funding development and testing.
- Access to "Stretch" goals as outlined below.
- Advance notification of bugs and security issues.
- Priority consideration of feature requests and bug reports.
- Private support and assistance via direct email.
The standard price for business support is $120.00 USD / year / production instance (including reserve instances). Please contact us for more details.
Each paying business customer is entitled to one vote to prioritize the below work.
- Add support stream prioritization in async-http.
- Add support for rolling restarts in async-container.
- Add support for hybrid process/thread model in async-container.
- Asynchronous Postgres and MySQL database adapters for ActiveRecord in async-postgres and async-mysql.
- Fork it
- Create your feature branch (
git checkout -b my-new-feature
) - Commit your changes (
git commit -am 'Add some feature'
) - Push to the branch (
git push origin my-new-feature
) - Create new Pull Request
If you want to donate to this project, you may do so using BitCoin. All money donated this way will be used to further development of this and related open source projects.
We take the security of our systems seriously, and we value input from the security community. The disclosure of security vulnerabilities helps us ensure the security and privacy of our users. If you believe you've found a security vulnerability in one of our products or platforms please tell us via email.
Released under the MIT license.
Copyright, 2018, by Samuel G. D. Williams.
Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions:
The above copyright notice and this permission notice shall be included in all copies or substantial portions of the Software.
THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.