Proxy over gRPC

pog-server is a HTTP proxy which uses gRPC for sending bytes:

User <-(HTTP proxying)-> pog client <-(gRPC)-> pog server <-(HTTP proxying)-> destination server

Статья на Хабре

Applications of Proxy over gRPC (PoG)

• terraform: if https://registry.terraform.io/ is
  not currently available in your region
  , one can use it with a proxy like:

• If a site is not available from your IP address, use the proxy, e.g. with Google Chrome:

Simple example of use

The pog server:

The pog client:

A user of the HTTP proxy:

Here the user reaches the destination URL ifconfig.me via the HTTP proxy at https://localhost:18080 .

How to build

Go programming language version > 1.21 is required.

How to build Docker image

How to deploy to Google Cloud Run using Terraform¹

The server part can be deployed into GCP Cloud Run, see

as an example. First, let's gate our server service with auth with login and password:

This JSON value we assign to the env variable POG_AUTH_ITEM1, see

Having a GCP project

, do:

Optionally, let's gate user requests to our client service, login=user and password=password:

The client part:

Finally, the user request:

Optional tweaks to GCP service config (

pog-server.tf

)

For a single home usage, the Cloud Run costs might be not affordable (around 15$/m), so one might try those tweaks:

Options

All the options are represented as environment variables.

The server part options:

The client part options:

The common options:

Metrics and operations

Both PoG server and client provides Prometheus metrics at

. An example:

What does it mean:

• rpcs
  : stats how much requests were processed and their success
• tunnelling_connections_total
  : a gauge featuring how many connections are being proccessed now; if the value is growing over time then there is a memory leak
• auth_item_earliest_expiry
  : time when a user account is to expire (both at server and client side)

Footnotes

1. One can deploy a Cloud Run service with the console, too. ↩