SQUID Proxy - How to Bypass proxy for specific URL
We are using Squid proxy in transparent mode. Our internal network subnet (10.0.0.0/24) has been segregated such that servers reside in (10.0.0.0/24) and all other users are 10.0.8-11.0/24. With this setup, I am able to configure Squid to Bypass the Proxy for servers.
We have a team of Web Developers who frequently test sites hosted in Amazon's Cloud. The IPs for these sites change quite frequently, which makes it difficult to bypass the proxy for destination IPs.
The "cache_deny" directive does not work for us either, as the users frequently "hack" their hosts files so that they can temporarily point a domain to a test server. When Doing this, it appears that squid continues to send the request to the original IP address, instead of the locally entered IP address.
so for example, if we have a cache deny for www.mytestdomain.com, which is hosted on a server at 220.127.116.11, and we are doing some testing, so hack the hosts file on the developers PC to point that domain to 18.104.22.168, the page that is returned from squid is the original at 22.214.171.124.
If the same tests are carried out from a server in the "Bypass Proxy for the following Source IPs", then this works ok. Similarly, entering the Destination IPs into the second bypass field works great, but with amazon IPs changing as frequently as they do, this is impractical.
Is there a way of bypassing the proxy altogether for a specific domain / url?
The "cache_deny" directive does not work for us either, as the users frequently "hack" their hosts files so that they can temporarily point a domain to a test server. […] Is there a way of bypassing the proxy altogether for a specific domain / url?
Unless I am misunderstanding your question, I think you've answered it yourself? The 'cache deny' directive is matching the domain you are entering. The difference is that pfSense is using DNS to look up the IP address while the end user is using their 'hosts' file (which pfSense has no knowledge of). The answer is to synchronize the DNS between pfSense and the end user so they both agree on the IP addresses. Of course this is normally done by telling the end user not to modify their hosts file otherwise they will experience caching. (Alternatively the end user could request that the DNS be modified. The key is that pfSense and the client need to be receiving the same IP address information.)
If the same tests are carried out from a server in the "Bypass Proxy for the following Source IPs", then this works ok.
If asking them not to modify the hosts file or modifying the DNS is impractical, then you already mentioned the next best solution which it to disable the proxy for those particular developers. (Are their IP addresses changing frequently too? If so, perhaps DDNS is called for?)
Thanks MindfulCoyote. You are correct. I am going to create a subnet specific for developers, and bypass the proxy altogether for them. Its the "least worst" solution on this occasion, but we lose the ability to track their behaviour which is a shame.