Selective web proxy command?

KendersPlace

Hello all. Got a question I've been working on for a week now and can't figure it out. 😕

Assume I have a proxy server running (squid). I can easily manually configure a web browser to use this proxy server. Now assume they are random visitors to a site. I want to push certain content through the proxy, but don't want the user to have to configure their browser for just that single purpose.

How can I selectively point a browser to a proxy either by adding something to the URL string or coding something in the PHP? It would probably have to execute client side as it has nothing to do with the connection with the web server showing the page.

More detail:

I have a room full of web servers behind a firewall. I can only get access to those boxes from ONE source (my proxy box). I want to route traffic to the target web servers but all traffic must pass the proxy and pull live from the target servers (I can't just replicate the content to the proxy as it is a real time stream, if that makes sense).

Help help help help???? This has become quite frustrating. Thanks much in advance.

tha_mink

I think you need to be more specific. What side of the proxy are the clients on? Both?

Sounds more like a squid issue than anything else. Can't you just configure squid to handle the decision to pull from the webserver or serve cached content? Why does any of this have anything to do with the client?

More info please.

KendersPlace

Thanks for the reply. Sorry if that was confusing. Here's a more broad picture. I think this may be my lack of knowledge of squid... but here's what I need to acomplish.

I have multiple remote servers. These remote servers take operating data from machinery in a plant (temp, speed, ect) and make it available as a stream from it's internal browser (very specialized product, but standard web server interface.

I have filed service people who log in from all over the country to view this live machine data. The goal is, the field rep will eventually get re-directed to the actual equipment monitor servers.

For security reasons, they only want to open up each remote equipment monitor to traffic from one IP. I would like to get squid to pull the live data stream and filter it through to the actual field reps (the network clients). We could easily do this by getting the client to change their browser properties to use the squid server. The company does not want to do that.

I need a way to pull a page from my web server (any web server for that matter)... that page then tells the client's browser to retrieve "http:xxx.xxx.xxx/index" page but to retrieve it through my squid server, not directly as it would do by default.

I hope that clears it up. I know this could be done with VPNs and some high end hardware on each end, but that's a lot of expense for them now, and it would be great if I could get it to run this way.

Thanks!

tha_mink

So it sounds like you have all these machines running webservers that serve out live data when requested. So you don't need squid at all really. You just need the gateway machine to be running a webserver and that webserver can request the pages from the internal servers. There are lots of ways to do this...fopen comes to mind.

I don't understand why you are running all this through squid when all you need is to have the gateway machine request data from the internal servers. Block all outside access to each internal machine, don't allow any communication to the outside world and control data flow through your code on the gateway server. Right?

So instead of having the client (your reps) request stuff through squid, you can have one web interface and control all data flow through your main gateway webserver.

(A better network design would be to have something like...
Firewall box>>DMZ>>gateway webserver>>internal network but still, since you can control internal network access through the main gateway webserver you should be pretty safe.)

KendersPlace

Once again, thank you. We actually considered this approach. This works fine if it is only text on the internal servers. Unfortunately, this stuff is so old that it pushes it's data in a non standard (non-ascii) form. The data flows to the viewer's browser via and activeX control (.ocx) sent to the viewing browser.

I put this in front of some developers I know and they couldn't figure out how the data was being streamed into the viewing component, so I don't think I can just re-create it.

The only solution that has worked so far is to do squid, then set the client's browser to use the squid proxy, then request the destination url, which is then routed (activeX, data, and all) back through squid.

I think fopen will only work with returned text as the result?

Thanks,
Kevin