Kage (kah-geh) is a shadow proxy server to duplex HTTP requests
Kage (kah-geh) is a shadow proxy server to duplex HTTP requests
Kage (kah-geh) is an HTTP shadow proxy server that sits between clients and your server(s) to enable "shadow requests".
Kage can be used to duplex requests to the master (production) server and shadow servers that have newer code changes that are going to be deployed. By shadowing requests to the new code you can make sure there are no big/surprising changes in the response in terms of data, performance and database loads etc.
Kage is built with EventMachine and em-proxy and all shadow requests are done asynchronously, while responses from master are sent back to the client without blocking the network, so clients will never notice any delays even when shadow traffic is made.
You can customize the behavior of Kage with simple callbacks, when it chooses which backends to send shadow requests to (or not at all), appends or deletes HTTP headers per backend, and examines the complete HTTP response (including headers and body).
Kage does not yet support:
…Read more sample code under the examples/ directory.
Kage supports keep-alives for single backend requests, i.e. for requests where on_select_backends returns only the master backend.
To make on_backend_finished callback simpler, if the current request matches with multiple backends, Kage sends Connection: close to the backends so that the callback will only get one response per backend in responses hash, which would look like:
responses = {
:original => {:data => "(RAW HTTP response)", :elapsed => 0.1234},
:sandbox => {:data => "(RAW HTTP response)", :elapsed => 0.2333},
}Tatsuhiko Miyagawa, Yusuke Mito
Ilya Grigorik, Jos Boumans
No open issues yet, or sync has not completed.