Here's my first-pass at a very basic connection scheduler:
There are two connection queues: eager and lazy (think of better names). All peers-to-be-dialed start in the eager queue and one peer is dialed every n seconds. If a connection attempt fails, the peer is moved to the lazy queue. One peer from the lazy queue is dialed every n x 10 seconds (or similar). If a connection attempt succeeds, the peer is moved back to the eager queue.
Combine (address, public_key) data from replication.toml with connection data passed via --connect CLI option
Here's my first-pass at a very basic connection scheduler:
There are two connection queues:
eagerandlazy(think of better names). All peers-to-be-dialed start in theeagerqueue and one peer is dialed everynseconds. If a connection attempt fails, the peer is moved to thelazyqueue. One peer from thelazyqueue is dialed everyn x 10seconds (or similar). If a connection attempt succeeds, the peer is moved back to theeagerqueue.(address, public_key)data fromreplication.tomlwith connection data passed via--connectCLI option