https://github.com/praveingk/serverless-relay
https://github.com/praveingk/serverless-relay
Last synced: 3 months ago
JSON representation
- Host: GitHub
- URL: https://github.com/praveingk/serverless-relay
- Owner: praveingk
- Created: 2023-10-16T20:46:22.000Z (over 1 year ago)
- Default Branch: main
- Last Pushed: 2023-10-27T18:33:21.000Z (over 1 year ago)
- Last Synced: 2025-01-31T04:33:04.431Z (4 months ago)
- Language: Go
- Size: 1.68 MB
- Stars: 0
- Watchers: 1
- Forks: 0
- Open Issues: 0
-
Metadata Files:
- Readme: README.md
Awesome Lists containing this project
README
# Serverless-relay
A Relay maintaining persistent connection between serverless functions.
A relay is needed when two clients want to send communicate between each other without having to host a server, and maintaining connections.
This is especially the case when we want one or more serverless functions/jobs to communicate with each other.
# Why is this needed?
A serverless job generally doesnt accept incoming connections. Hence, when two serverless jobs needs to communicate with each other, they would need to connect through a central relay/router.
In the context of Clusterlink, currently ClusterLink gateways expect reachability to the destination service which has to receive connections. An application if it wants to connect to a service, it will connect to the GWs and the GW in the destination cluster needs to make a tcp connection with the destination service.
However, when two transient jobs/functions running in serverless need to contact each other over a tcp connection, there is no generic way of doing this assuming the functions do not have/want to expose publicly reachable endpoints. This problem is currently not solved by Clusterlink, since it requires the GW to initiate a connection with one of the functions, which requires at least one of the functions to have a reachable endpoint accepting connections.
# How does serverless-relay solve this?
The serverless-relay runs as a separate process within a VM.
It does the following:
1) Binds and listen to a specific port, for the job/function to connect and send messages to the remote service. The remote service is already discovered by the gateway.
2) The relay at one end accepts connection from a func1, and other end accepts connection from func2. Ideally it should validate the func1 & func2 using their certificates.
3) When both funcs connect to the relay, an end-to-end connection is established, and funcs upgrade the connection to mTLS, and communicate with each other.
Details :
1) The relay starts a listener for each {func1, func2} pair communication.
2) It verifies the functions and manages routing between them.
# Steps to run serverless-relay
make build
./bin/frelay start --port
# Run a test client
export MODE=flock_tls_server
./bin/client_function
# Run a test server
export MODE=flock_tls_client
./bin/client_function
## Problem
Forwarding over the TCP socket of the tls connection fails.
## TODO
1) Move relay configuration to separate API, e.g. specify addRoute(func1, func2), and deleteRoute(func1,func2).