2y ago

Selfhosting a proxy at home?

Hello! I have a server that runs 24/7, and have recently started doing some stuff that requires scraping the web. The websites are detecting the server’s IP to not be residential though, and it’s causing issues.

I’d like to host a proxy server on the small server I have running 24/7 in my house, so that everything for that 1 page could be proxied through it. Does anyone have any idea how I’d set up a server like that? Thanks.

17 comments

Max-P already provided good options, but I have to ask what I, and probably other people, wonder : why don't you just run that scrapping program from your home server, then?
- The scraping program saves large files which I don’t have space for on the tiny raspi
  
  (sorry for the double post, the instance I'm on was throwing errors)
  Gotcha, thanks for satisfying my curiosity. :) Of course, you can plug a usb drive on the Pi, but you know better what your needs are. Good luck!
How much are you scraping? You may end up getting your home up blocked.
- Nah, it’s not much. Maybe 10 pages a day.
You can pretty easily install Squid, it's fairly simple to configure and works well for most use cases. Just a plain simple HTTP proxy.
You could also set up a VPN to your home to achieve something similar, by binding some requests to the VPN IP. It's a bit harder to set up however as it involves routing tables, route metrics and conditionally binding the outgoing connection to a specific interface
- Thanks, sounds like Squid will be perfect. I’ll just need to figure out some way to connect. I wish I could just open a port, but it hasn’t been working since I enabled IPv6 on my router. Do you think I could make it accessible through cloudflare tunnels?
  
  Probably not, look into wireguard or tailscale.
  
  Cloudflare tunnels won't work as Cloudflare won't tunnel HTTP proxy traffic, at least as far as I know.
  What you can do however is have your home server VPN into your remote server, then your remote server will have no problem connecting to Squid over the VPN link. WireGuard is very simple to configure like that, probably 5-10 lines of config on each end. You don't need any routing or forwarding or anything, just a plain VPN with 2 peers that can ping eachother, so no ip_forward or iptables -j MASQUERADE needed or anything that most guides would include. You can also use something like Tailscale, anything that will let the two machines talk to eachother.
  Depending on the performance and reliability needs, you could even just forward a port with SSH. Connect to your remote server from the home server with something like ssh -N -R localhost:8088:localhost:8080 $remoteServer and port 8088 on the remote will forward to port 8080 on the home server as long as that SSH connection is up. -N simply makes SSH not open a shell on the remote, dedicating the SSH session to the forwarding. Nice and easy, especially for prototyping.
Besides other answers - you could just use SSH port forwarding. Remote would be your home server, local would be your "cloud" server. You should initiate the connection from the cloud server to your home server. Playing with local ports would enable you filter what domains are used for the proxy.
I rarely use it, so the exact syntax is gone from my memory. It is a bit tricky at first, but definitely not rocket science to figure out.
Once the connection is established - you would point your scraper to https://localhost:localport
- This is the most simple solution and probably a lot safer than the alternatives. Another good option would be to use OpenVPN.
I ended up hosting the scraper service on my home server and exposing it through Cloudflare Tunnels, as the service is pretty much just an API that doesn’t really do much data downloads

17 comments