We have a lot of ad-based traffic that pass various query string params on the URL, but these do not affect the content/output of the pages.
I currently have the following nginx proxy cache configuration, which passes uncached requests to another backend server that actually returns the content.
proxy_cache_key "$request_method@$scheme://$host:$server_port$uri"; proxy_cache_valid 200 15m; proxy_pass http://backend;
I am just using the
$uri and not the
$args within the cache key so that both of the following would pull from the same cache, which is working great
But, if I start caching the 301 redirect responses from the backend server
proxy_cache_valid 301 15m;
Then we start caching the 301’s returned from the backend server, using the non-arg cache key which causes a problem because:
the first request to http://somedomain.com/?aid=129f58ad4af8f9de08bbd6bb7df22850 redirects to... http://www.somedomain.com/?aid=129f58ad4af8f9de08bbd6bb7df22850 but http://somedomain.com/ or http://somedomain.com/?foo=bar pulls from the cache and redirects to... http://www.somedomain.com/?aid=129f58ad4af8f9de08bbd6bb7df22850
Is there a way to have the 301 redirects cached that ignore the querystring and just passthrough whatever querystring params are present on the request?
Based on the semantics of what these URLs do, I think it would be best if nginx did not attempt to cache these requests at all; rather, to always pass them up to your application, so that you can do the appropriate tracking.
Fortunately this is a one-liner:
will bypass the nginx cache and pass the request up to your app if the
aid argument is present in the query string.
This work is licensed under a Creative Commons Attribution-ShareAlike 3.0 Unported License.