goose 0.17.0

A load testing framework inspired by Locust.
Documentation 
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70

use tokio::time;

use crate::util;

/// This throttle thread limits the maximum number of requests that can be made across
/// all GooseUser threads. When enabled, GooseUser threads must add a token to the
/// bounded channel before making a request, and this thread limits how frequently
/// tokens are removed thereby throttling how fast requests can be made. It is an
/// implementation of the leaky bucket algorithm as a queue: instead of leaking the
/// overflow we asynchronously block. More information on the leaky bucket algorithm
/// can be found at: https://en.wikipedia.org/wiki/Leaky_bucket
pub async fn throttle_main(
    throttle_requests: usize,
    throttle_receiver: flume::Receiver<bool>,
    parent_receiver: flume::Receiver<bool>,
) {
    // Use microseconds to allow configurations up to 1,000,000 requests per second.
    let mut sleep_duration = time::Duration::from_micros(1_000_000 / throttle_requests as u64);
    let tokens_per_duration;

    let ten_milliseconds = time::Duration::from_millis(10);
    debug!(
        "sleep_duration: {:?} ten_milliseconds: {:?}",
        sleep_duration, ten_milliseconds
    );

    // Keep sleep_duration at least ~10ms as `delay_for` has millisecond granularity.
    if sleep_duration < ten_milliseconds {
        tokens_per_duration = (ten_milliseconds.as_nanos() / sleep_duration.as_nanos()) as u32;
        sleep_duration *= tokens_per_duration;
    } else {
        tokens_per_duration = 1;
    }

    info!(
        "throttle allowing {} request(s) every {:?}",
        tokens_per_duration, sleep_duration
    );

    // One or more token gets removed from the throttle_receiver bucket at regular
    // intervals. The throttle_drift variable tracks how much time is spent on
    // everything else, and is subtracted from the time spent sleeping.
    let mut throttle_drift = tokio::time::Instant::now();

    // Loop and remove tokens from channel at controlled rate until load test ends.
    loop {
        debug!(
            "throttle removing {} token(s) from channel",
            tokens_per_duration
        );
        throttle_drift = util::sleep_minus_drift(sleep_duration, throttle_drift).await;

        // A message will be received when the load test is over.
        if parent_receiver.try_recv().is_ok() {
            // Close throttle channel to prevent any further requests.
            info!("load test complete, closing throttle channel");
            drop(throttle_receiver);
            break;
        }

        // Remove tokens from the channel, freeing spots for request to be made.
        for token in 0..tokens_per_duration {
            // If the channel is empty, we will get an error, so stop trying to remove tokens.
            if throttle_receiver.try_recv().is_err() {
                debug!("empty channel, exit after removing {} tokens", token);
                break;
            }
        }
    }
}