Silly simple cave man here can't make things work

[shoutmarble]

How do I make a reverse proxy with g3?

Your documentation is very good but my cave man literacy level is lower than any other recorded caveman.

I understand some caveman technology and I can often make other rocks that cavemen use work. I usually see how the other cavemen smash rocks and then repeat their rock smashing actions with my own rocks.

But these rocks look a little different than what this caveman encountered seeing other caveman use and is not getting an immediate fire in his cave.

How do I smash these rocks together to make a reverse proxy?

Caveman first procured the rocks like so.

docker build -f g3proxy/docker/alpine.Dockerfile github.com/bytedance/g3 -t g3proxy:prox

And this Caveman then sees that his rocks are indeed in his cave now.

root@server:~/g3# docker images
REPOSITORY       TAG       IMAGE ID       CREATED        SIZE
g3proxy          prox      2be73601a571   11 hours ago   39.6MB

Caveman then poked the rock and not smash the rock yet to verify it is a working rock.

root@server:~/g3# docker run g3proxy:prox
g3proxy 1.11.2
Features: c-ares hickory-dns quinn
OpenSSL Variant: default
Rustls Provider: ring
Compiler: rustc 1.84.0 (9fc6b4312 2025-01-07) (Stable)
Host: x86_64-unknown-linux-musl, Target: x86_64-unknown-linux-musl
Profile: release, Opt Level: 3, Debug: false

Caveman now wants to use his rock on his VPS I mean Cave to point from the entrance of his cave to another
special rock in his cave whenever someone enters his Cave.

root@server:~/g3# cat whoami.yml

services:
    whoami:
        image: traefik/whoami

Caveman is not good with "words" and wants to see rocks!

I think this is the cave pit that I need to use for the "whoami" reverse-proxy

https://github.com/bytedance/g3/tree/master/g3proxy/examples/simple_http_proxy

[zh-jq]

You need to use https://github.com/bytedance/g3/blob/master/g3proxy/examples/simple_http_gateway/g3proxy.yaml this as an example HTTP reverse proxy config. The http_rproxy server is really simple implementation of HTTP reverse proxy compared to others such as nginx/envoy, so I'm not sure whether it will fit your needs.

You need to mount your g3proxy.yaml config file or config dir when start the docker container, and change the entry command to g3proxy -c /path/to/g3proxy.yaml. Or you can build a new image by adding the config file and update the default RUN command. Anyway I'm not good at docker usage, so you have to figure out the best way yourself.

[shoutmarble]

root@server:~/g3# docker run -v ./g3:/g3_http g3proxy:prox -c /g3_http/g3proxy/examples/simple_http_proxy/g3who.yaml

root@server:~/g3# pwd
/root/g3

root@server:~/g3# ll
total 24
drwxr-xr-x  3 root root 4096 Jan 12 04:30 ./
drwx------ 17 root root 4096 Jan 12 00:44 ../
drwxr-xr-x 17 root root 4096 Jan 11 11:17 g3/
-rwxr-xr-x  1 root root  210 Jan 12 04:30 rm-all.sh*
-rwxr-xr-x  1 root root   28 Jan 12 04:30 stop.sh*
-rw-r--r--  1 root root   64 Jan 12 04:18 whoami.yml

root@server:~/g3# tree g3/g3proxy/examples/simple_http_proxy
g3/g3proxy/examples/simple_http_proxy
├── cert.pem
├── g3proxy.yaml
├── g3who.yaml
└── key.pem

1 directory, 4 files

whoami.yml (I did not put in ports: "80:80")

root@server:~/g3# cat whoami.yml

services:
    whoami:
        image: traefik/whoami

root@server:~/g3# docker compose -f whoami.yml up
[+] Running 4/4
 ✔ whoami Pulled                                                         2.1s
   ✔ 0b8c4591162f Pull complete                                          0.8s
   ✔ 733db08f86a6 Pull complete                                          0.8s
   ✔ a5d67c72e18d Pull complete                                          0.9s
[+] Running 2/2
 ✔ Network g3_default     Created                                        0.1s
 ✔ Container g3-whoami-1  Created                                        0.1s
Attaching to whoami-1
whoami-1  | 2025/01/12 04:49:03 Starting up on port 80

root@server:~/g3# docker ps
CONTAINER ID   IMAGE            COMMAND                  CREATED          STATUS          PORTS     NAMES
d32ce89fd017   g3proxy:prox     "/usr/bin/g3proxy -c…"   11 seconds ago   Up 10 seconds             epic_cray
570828d34caa   traefik/whoami   "/whoami"                32 minutes ago   Up 32 minutes   80/tcp    g3-whoami-1

root@server:~/g3# curl localhost:80
curl: (7) Failed to connect to localhost port 80 after 3 ms: Couldn't connect to server

       root@server:~/g3# cat g3/g3proxy/examples/simple_http_proxy/g3who.yaml
       ---
       runtime:
         thread_number: 2
       worker:
         sched_affinity: true
       log: journal
       controller:
         local:
           recv_timeout: 30
           send_timeout: 1
       server:
         - name: test
           escaper: default
           type: http_proxy
           listen:
             address: "[::]:10086"
           tls_client: {}
           tcp_sock_speed_limit: 100M
         - name: test-tls
           tls_server:
             certificate: cert.pem
             private_key: key.pem
           escaper: default
           type: http_proxy
           listen:                              +--------------------------------------+
             address: "[::]:10087"              | How do I let g3proxy know about the  |
           tcp_sock_speed_limit: 100M           | name, ip, and port of my whoami.yml? |
         - name: test-alias                     +--------------------------------------+
           type: plain_tcp_port
           listen:                              +------------------------------------------------------+
             address: "[::]:80"   <------+      | -rw-r--r+-  1 root root   64 Jan 12 04:18 whoami.yml |
           server: test                  |      | root@server:~/g3# cat whoami.yml                     |
       resolver:                         |      +------------------------------------------------------+
         - name: default                 |      +--------------------------------+
           type: c-ares                  |      | services:                      |
           server: 127.0.0.1             |      |     whoami:                    |
       escaper:                          +------+         image: traefik/whoami  |
         - name: default                        |                                |
           type: direct_fixed                   +--------------------------------+
           no_ip^6: true
           resolver: default
           resolve_strategy: IPv4Only
           tcp_sock_speed_limit: 80M
           udp_sock_speed_limit: 10M
           egress_network_filter:
             default: allow
             allow:
               - 127.0.0.1
       root@server:~/g3#

              

[zh-jq]

The g3proxy container has a http_proxy server, which is a forward proxy mode server, you can test it with curl -x http://<g3proxy-container-ip>:<port> http://<whoami-container-ip>:<port>.

I think you need to expose the ports in g3proxy container to the host, so you can run it from the host.
But I don't known how to access to the whoami container from the g3proxy container, it depends on the network config I think.

I will try this when I have some free time.

[shoutmarble]

I checked out https://copilot.microsoft.com/, and it appears to have enough information to
point me in the right direction.

I will test something from chatgpt like shown below in a little bit.

Thank you for the information regarding curl and configuring up a network.

I can't wait to test that out.

g3proxy.yaml

runtime:
  thread_number: 2

log: journal

controller:
  local:
    recv_timeout: 30
    send_timeout: 1

server:
  name: test
  escaper: default
  type: http_rproxy
  listen: [::]:10080
  conn_limit: 100M
  hosts:
    - exact_match: my-web-app
      upstream: my-web-app:80
  resolver:
    name: default
    type: c-ares
    server: 223.5.5.5
  escaper: []
  no_ipv6: true
  resolve_strategy: IPv4Only
  tcp_conn_limit: 80M
  udp_relay_limit: 10M
  egress_network_filter: []

docker network

docker network create my_network

docker web-app

docker run -d --name my-web-app --network my_network my-web-app-image

docker run g3proxy

docker run -d --name g3proxy --network my_network -v /path/to/g3proxy.yaml:/g3proxy.yaml bytedance/g3proxy

[zh-jq-b]

The g3proxy.yaml file can be changed to:

log: stdout

resolver:
  - name: default
    type: c-ares
    server: 127.0.0.1

escaper:
  - name: default
    type: direct_fixed
    no_ipv6: true
    resolve_strategy: IPv4Only

server:
  - name: test
    escaper: default
    type: http_rproxy
    listen: [::]:8080
    hosts:
      - exact_match: my-web-app.local
        upstream: my-web-app.local:80
    resolver: default
    escaper: default

Make sure the DNS server set in resolver section match this https://docs.docker.com/engine/network/#dns-services, and the hostname set in the server/hosts should be resolvable by that DNS server.