In 0.2 release, with server side flux implementation, we went multiprocess, i.e., separate store, bouncer, cli and web processes. With the basic server side flux framework in place, it's time to go multiple bouncer instances this time. The goal is to:
distribute IRC bot instances across multiple bouncer instances
to be fault tolerant, i.e., if a bouncer instance dies, the system must be able to re create the deceased bots in the failed bouncer in the current running bouncers
implement load balancer for the bouncers
TODOS:
[ ] Implement dynamic bouncer registration and discovery.
[ ] Implement a strategy to distribute IRC bots across multiple bouncers (Round Robin would work for now)
[ ] Implement a registry for IRC bots running in bouncers
[ ] When a bouncer dies, distribute bots running on it across working bouncers and connect to IRC
In 0.2 release, with server side flux implementation, we went multiprocess, i.e., separate
store,bouncer,cliandwebprocesses. With the basic server side flux framework in place, it's time to go multiple bouncer instances this time. The goal is to:TODOS:
005 RPL_BOUNCEIRC message as mentioned in https://tools.ietf.org/html/rfc2812Let's create separate issues for each of the above TODOs and start hacking