tablejohn/src/worker/coordinator.rs

104 lines
3.2 KiB
Rust

//! Coordinate performing runs across servers.
use std::time::Duration;
use time::OffsetDateTime;
use tokio::sync::mpsc;
struct Server {
name: String,
poke: mpsc::UnboundedSender<()>,
}
pub struct Coordinator {
servers: Vec<Server>,
active: usize,
active_since: OffsetDateTime,
busy: bool,
}
impl Coordinator {
pub fn new() -> Self {
Self {
servers: vec![],
active: 0,
active_since: OffsetDateTime::now_utc(),
busy: false,
}
}
pub fn register(&mut self, name: String, poke: mpsc::UnboundedSender<()>) {
// TODO Assert that no duplicate names exist?
self.servers.push(Server { name, poke });
}
pub fn active(&self, name: &str) -> ActiveInfo {
let active_server = self.servers.get(self.active);
let active = active_server.filter(|s| s.name == name).is_some();
ActiveInfo {
active,
active_since: self.active_since,
busy: self.busy,
}
}
pub fn look_busy(&mut self, name: &str) {
// Check just to prevent weird shenanigans
if !self.active(name).active {
return;
}
self.busy = true;
}
pub fn move_to_next_server(&mut self, name: &str) {
// Check just to prevent weird shenanigans
if !self.active(name).active {
return;
}
// At least one server (the current one) must be registered according to
// the previous check
assert!(!self.servers.is_empty());
self.active += 1;
self.active %= self.servers.len();
self.active_since = OffsetDateTime::now_utc();
self.busy = false;
// When the worker seeks work and a queue is idle, the next server
// should be queried immediately. Otherwise, we'd introduce lots of
// delay in the multi-server case were most queues are empty.
//
// However, if all server's queues were empty, this would generate a
// slippery cycle of requests that the worker sends as quickly as
// possible, only limited by the roundtrip time. Because we don't want
// this, we let the first task wait its full timeout. Effectively, this
// results in iterations starting at least the ping delay apart, which
// is pretty much what we want.
//
// The way this is implemented currently is sub-optimal however: If the
// chain takes even a fraction longer than the previous iteration, tasks
// will send two requests back-to-back: The first because their ping
// timeout ran out, and the second because they were poked. So far, I
// haven't been able to think of an elegant solution for this.
if self.active > 0 {
let _ = self.servers[self.active].poke.send(());
}
}
}
#[derive(Clone, Copy)]
pub struct ActiveInfo {
pub active: bool,
pub active_since: OffsetDateTime,
pub busy: bool,
}
impl ActiveInfo {
pub fn in_batch(&self, batch_duration: Duration) -> bool {
let batch_end = self.active_since + batch_duration;
let now = OffsetDateTime::now_utc();
now <= batch_end
}
}