Infinite unbound loop in analytics processor lead to 100% CPU usage

[Totodore]

The following code loops indefinitely without pause, this leads to 100% CPU usage in all our environment (dev/qual/prod). We disabled the analytics processor in the meantime.

Solution

Use rx.recv_timeout rather than try_recv which is non blocking.

flagsmith-rust-client/src/flagsmith/analytics.rs

Lines 38 to 61 in 693b1fe

	.spawn(move || {
	let mut last_flushed = chrono::Utc::now();
	loop {
	let data = rx.try_recv();
	let mut analytics_data = analytics_data_locked.write().unwrap();
	match data {
	// Update the analytics data with feature_id received
	Ok(feature_name) => {
	analytics_data
	.entry(feature_name)
	.and_modify(|e| *e += 1)
	.or_insert(1);
	}
	Err(flume::TryRecvError::Empty) => {}
	Err(flume::TryRecvError::Disconnected) => {
	debug!("Shutting down analytics thread ");
	break;
	}
	};
	if (chrono::Utc::now() - last_flushed).num_milliseconds() > timer as i64 {
	flush(&client, &analytics_data, &analytics_endpoint);
	analytics_data.clear();
	last_flushed = chrono::Utc::now();
	}