Line data Source code
1 : use std::borrow::Cow;
2 : use std::collections::hash_map::RandomState;
3 : use std::hash::{BuildHasher, Hash};
4 : use std::sync::atomic::{AtomicUsize, Ordering};
5 : use std::sync::Mutex;
6 :
7 : use anyhow::bail;
8 : use dashmap::DashMap;
9 : use itertools::Itertools;
10 : use rand::rngs::StdRng;
11 : use rand::{Rng, SeedableRng};
12 : use tokio::time::{Duration, Instant};
13 : use tracing::info;
14 :
15 : use crate::ext::LockExt;
16 : use crate::intern::EndpointIdInt;
17 :
18 : pub struct GlobalRateLimiter {
19 : data: Vec<RateBucket>,
20 : info: Vec<RateBucketInfo>,
21 : }
22 :
23 : impl GlobalRateLimiter {
24 0 : pub fn new(info: Vec<RateBucketInfo>) -> Self {
25 0 : Self {
26 0 : data: vec![
27 0 : RateBucket {
28 0 : start: Instant::now(),
29 0 : count: 0,
30 0 : };
31 0 : info.len()
32 0 : ],
33 0 : info,
34 0 : }
35 0 : }
36 :
37 : /// Check that number of connections is below `max_rps` rps.
38 0 : pub fn check(&mut self) -> bool {
39 0 : let now = Instant::now();
40 0 :
41 0 : let should_allow_request = self
42 0 : .data
43 0 : .iter_mut()
44 0 : .zip(&self.info)
45 0 : .all(|(bucket, info)| bucket.should_allow_request(info, now, 1));
46 0 :
47 0 : if should_allow_request {
48 0 : // only increment the bucket counts if the request will actually be accepted
49 0 : self.data.iter_mut().for_each(|b| b.inc(1));
50 0 : }
51 :
52 0 : should_allow_request
53 0 : }
54 : }
55 :
56 : // Simple per-endpoint rate limiter.
57 : //
58 : // Check that number of connections to the endpoint is below `max_rps` rps.
59 : // Purposefully ignore user name and database name as clients can reconnect
60 : // with different names, so we'll end up sending some http requests to
61 : // the control plane.
62 : pub type WakeComputeRateLimiter = BucketRateLimiter<EndpointIdInt, StdRng, RandomState>;
63 :
64 : pub struct BucketRateLimiter<Key, Rand = StdRng, Hasher = RandomState> {
65 : map: DashMap<Key, Vec<RateBucket>, Hasher>,
66 : info: Cow<'static, [RateBucketInfo]>,
67 : access_count: AtomicUsize,
68 : rand: Mutex<Rand>,
69 : }
70 :
71 : #[derive(Clone, Copy)]
72 : struct RateBucket {
73 : start: Instant,
74 : count: u32,
75 : }
76 :
77 : impl RateBucket {
78 3000915 : fn should_allow_request(&mut self, info: &RateBucketInfo, now: Instant, n: u32) -> bool {
79 3000915 : if now - self.start < info.interval {
80 3000907 : self.count + n <= info.max_rpi
81 : } else {
82 : // bucket expired, reset
83 8 : self.count = 0;
84 8 : self.start = now;
85 8 :
86 8 : true
87 : }
88 3000915 : }
89 :
90 3000909 : fn inc(&mut self, n: u32) {
91 3000909 : self.count += n;
92 3000909 : }
93 : }
94 :
95 : #[derive(Clone, Copy, PartialEq)]
96 : pub struct RateBucketInfo {
97 : pub(crate) interval: Duration,
98 : // requests per interval
99 : pub(crate) max_rpi: u32,
100 : }
101 :
102 : impl std::fmt::Display for RateBucketInfo {
103 19 : fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
104 19 : let rps = self.rps().floor() as u64;
105 19 : write!(f, "{rps}@{}", humantime::format_duration(self.interval))
106 19 : }
107 : }
108 :
109 : impl std::fmt::Debug for RateBucketInfo {
110 0 : fn fmt(&self, f: &mut std::fmt::Formatter<'_>) -> std::fmt::Result {
111 0 : write!(f, "{self}")
112 0 : }
113 : }
114 :
115 : impl std::str::FromStr for RateBucketInfo {
116 : type Err = anyhow::Error;
117 :
118 20 : fn from_str(s: &str) -> Result<Self, Self::Err> {
119 20 : let Some((max_rps, interval)) = s.split_once('@') else {
120 0 : bail!("invalid rate info")
121 : };
122 20 : let max_rps = max_rps.parse()?;
123 20 : let interval = humantime::parse_duration(interval)?;
124 20 : Ok(Self::new(max_rps, interval))
125 20 : }
126 : }
127 :
128 : impl RateBucketInfo {
129 : pub const DEFAULT_SET: [Self; 3] = [
130 : Self::new(300, Duration::from_secs(1)),
131 : Self::new(200, Duration::from_secs(60)),
132 : Self::new(100, Duration::from_secs(600)),
133 : ];
134 :
135 : pub const DEFAULT_ENDPOINT_SET: [Self; 3] = [
136 : Self::new(500, Duration::from_secs(1)),
137 : Self::new(300, Duration::from_secs(60)),
138 : Self::new(200, Duration::from_secs(600)),
139 : ];
140 :
141 : /// All of these are per endpoint-maskedip pair.
142 : /// Context: 4096 rounds of pbkdf2 take about 1ms of cpu time to execute (1 milli-cpu-second or 1mcpus).
143 : ///
144 : /// First bucket: 1000mcpus total per endpoint-ip pair
145 : /// * 4096000 requests per second with 1 hash rounds.
146 : /// * 1000 requests per second with 4096 hash rounds.
147 : /// * 6.8 requests per second with 600000 hash rounds.
148 : pub const DEFAULT_AUTH_SET: [Self; 3] = [
149 : Self::new(1000 * 4096, Duration::from_secs(1)),
150 : Self::new(600 * 4096, Duration::from_secs(60)),
151 : Self::new(300 * 4096, Duration::from_secs(600)),
152 : ];
153 :
154 19 : pub fn rps(&self) -> f64 {
155 19 : (self.max_rpi as f64) / self.interval.as_secs_f64()
156 19 : }
157 :
158 3 : pub fn validate(info: &mut [Self]) -> anyhow::Result<()> {
159 8 : info.sort_unstable_by_key(|info| info.interval);
160 3 : let invalid = info
161 3 : .iter()
162 3 : .tuple_windows()
163 4 : .find(|(a, b)| a.max_rpi > b.max_rpi);
164 3 : if let Some((a, b)) = invalid {
165 1 : bail!(
166 1 : "invalid bucket RPS limits. {b} allows fewer requests per bucket than {a} ({} vs {})",
167 1 : b.max_rpi,
168 1 : a.max_rpi,
169 1 : );
170 2 : }
171 2 :
172 2 : Ok(())
173 3 : }
174 :
175 24 : pub const fn new(max_rps: u32, interval: Duration) -> Self {
176 24 : Self {
177 24 : interval,
178 24 : max_rpi: ((max_rps as u64) * (interval.as_millis() as u64) / 1000) as u32,
179 24 : }
180 24 : }
181 : }
182 :
183 : impl<K: Hash + Eq> BucketRateLimiter<K> {
184 4 : pub fn new(info: impl Into<Cow<'static, [RateBucketInfo]>>) -> Self {
185 4 : Self::new_with_rand_and_hasher(info, StdRng::from_entropy(), RandomState::new())
186 4 : }
187 : }
188 :
189 : impl<K: Hash + Eq, R: Rng, S: BuildHasher + Clone> BucketRateLimiter<K, R, S> {
190 5 : fn new_with_rand_and_hasher(
191 5 : info: impl Into<Cow<'static, [RateBucketInfo]>>,
192 5 : rand: R,
193 5 : hasher: S,
194 5 : ) -> Self {
195 5 : let info = info.into();
196 5 : info!(buckets = ?info, "endpoint rate limiter");
197 5 : Self {
198 5 : info,
199 5 : map: DashMap::with_hasher_and_shard_amount(hasher, 64),
200 5 : access_count: AtomicUsize::new(1), // start from 1 to avoid GC on the first request
201 5 : rand: Mutex::new(rand),
202 5 : }
203 5 : }
204 :
205 : /// Check that number of connections to the endpoint is below `max_rps` rps.
206 1000457 : pub(crate) fn check(&self, key: K, n: u32) -> bool {
207 1000457 : // do a partial GC every 2k requests. This cleans up ~ 1/64th of the map.
208 1000457 : // worst case memory usage is about:
209 1000457 : // = 2 * 2048 * 64 * (48B + 72B)
210 1000457 : // = 30MB
211 1000457 : if self.access_count.fetch_add(1, Ordering::AcqRel) % 2048 == 0 {
212 488 : self.do_gc();
213 999969 : }
214 :
215 1000457 : let now = Instant::now();
216 1000457 : let mut entry = self.map.entry(key).or_insert_with(|| {
217 1000004 : vec![
218 1000004 : RateBucket {
219 1000004 : start: now,
220 1000004 : count: 0,
221 1000004 : };
222 1000004 : self.info.len()
223 1000004 : ]
224 1000457 : });
225 1000457 :
226 1000457 : let should_allow_request = entry
227 1000457 : .iter_mut()
228 1000457 : .zip(&*self.info)
229 3000915 : .all(|(bucket, info)| bucket.should_allow_request(info, now, n));
230 1000457 :
231 1000457 : if should_allow_request {
232 1000453 : // only increment the bucket counts if the request will actually be accepted
233 3000909 : entry.iter_mut().for_each(|b| b.inc(n));
234 1000453 : }
235 :
236 1000457 : should_allow_request
237 1000457 : }
238 :
239 : /// Clean the map. Simple strategy: remove all entries in a random shard.
240 : /// At worst, we'll double the effective max_rps during the cleanup.
241 : /// But that way deletion does not aquire mutex on each entry access.
242 488 : pub(crate) fn do_gc(&self) {
243 488 : info!(
244 0 : "cleaning up bucket rate limiter, current size = {}",
245 0 : self.map.len()
246 : );
247 488 : let n = self.map.shards().len();
248 488 : // this lock is ok as the periodic cycle of do_gc makes this very unlikely to collide
249 488 : // (impossible, infact, unless we have 2048 threads)
250 488 : let shard = self.rand.lock_propagate_poison().gen_range(0..n);
251 488 : self.map.shards()[shard].write().clear();
252 488 : }
253 : }
254 :
255 : #[cfg(test)]
256 : #[expect(clippy::unwrap_used)]
257 : mod tests {
258 : use std::hash::BuildHasherDefault;
259 : use std::time::Duration;
260 :
261 : use rand::SeedableRng;
262 : use rustc_hash::FxHasher;
263 : use tokio::time;
264 :
265 : use super::{BucketRateLimiter, WakeComputeRateLimiter};
266 : use crate::intern::EndpointIdInt;
267 : use crate::rate_limiter::RateBucketInfo;
268 : use crate::types::EndpointId;
269 :
270 : #[test]
271 1 : fn rate_bucket_rpi() {
272 1 : let rate_bucket = RateBucketInfo::new(50, Duration::from_secs(5));
273 1 : assert_eq!(rate_bucket.max_rpi, 50 * 5);
274 :
275 1 : let rate_bucket = RateBucketInfo::new(50, Duration::from_millis(500));
276 1 : assert_eq!(rate_bucket.max_rpi, 50 / 2);
277 1 : }
278 :
279 : #[test]
280 1 : fn rate_bucket_parse() {
281 1 : let rate_bucket: RateBucketInfo = "100@10s".parse().unwrap();
282 1 : assert_eq!(rate_bucket.interval, Duration::from_secs(10));
283 1 : assert_eq!(rate_bucket.max_rpi, 100 * 10);
284 1 : assert_eq!(rate_bucket.to_string(), "100@10s");
285 :
286 1 : let rate_bucket: RateBucketInfo = "100@1m".parse().unwrap();
287 1 : assert_eq!(rate_bucket.interval, Duration::from_secs(60));
288 1 : assert_eq!(rate_bucket.max_rpi, 100 * 60);
289 1 : assert_eq!(rate_bucket.to_string(), "100@1m");
290 1 : }
291 :
292 : #[test]
293 1 : fn default_rate_buckets() {
294 1 : let mut defaults = RateBucketInfo::DEFAULT_SET;
295 1 : RateBucketInfo::validate(&mut defaults[..]).unwrap();
296 1 : }
297 :
298 : #[test]
299 : #[should_panic = "invalid bucket RPS limits. 10@10s allows fewer requests per bucket than 300@1s (100 vs 300)"]
300 1 : fn rate_buckets_validate() {
301 1 : let mut rates: Vec<RateBucketInfo> = ["300@1s", "10@10s"]
302 1 : .into_iter()
303 2 : .map(|s| s.parse().unwrap())
304 1 : .collect();
305 1 : RateBucketInfo::validate(&mut rates).unwrap();
306 1 : }
307 :
308 : #[tokio::test]
309 1 : async fn test_rate_limits() {
310 1 : let mut rates: Vec<RateBucketInfo> = ["100@1s", "20@30s"]
311 1 : .into_iter()
312 2 : .map(|s| s.parse().unwrap())
313 1 : .collect();
314 1 : RateBucketInfo::validate(&mut rates).unwrap();
315 1 : let limiter = WakeComputeRateLimiter::new(rates);
316 1 :
317 1 : let endpoint = EndpointId::from("ep-my-endpoint-1234");
318 1 : let endpoint = EndpointIdInt::from(endpoint);
319 1 :
320 1 : time::pause();
321 1 :
322 101 : for _ in 0..100 {
323 100 : assert!(limiter.check(endpoint, 1));
324 1 : }
325 1 : // more connections fail
326 1 : assert!(!limiter.check(endpoint, 1));
327 1 :
328 1 : // fail even after 500ms as it's in the same bucket
329 1 : time::advance(time::Duration::from_millis(500)).await;
330 1 : assert!(!limiter.check(endpoint, 1));
331 1 :
332 1 : // after a full 1s, 100 requests are allowed again
333 1 : time::advance(time::Duration::from_millis(500)).await;
334 6 : for _ in 1..6 {
335 255 : for _ in 0..50 {
336 250 : assert!(limiter.check(endpoint, 2));
337 1 : }
338 5 : time::advance(time::Duration::from_millis(1000)).await;
339 1 : }
340 1 :
341 1 : // more connections after 600 will exceed the 20rps@30s limit
342 1 : assert!(!limiter.check(endpoint, 1));
343 1 :
344 1 : // will still fail before the 30 second limit
345 1 : time::advance(time::Duration::from_millis(30_000 - 6_000 - 1)).await;
346 1 : assert!(!limiter.check(endpoint, 1));
347 1 :
348 1 : // after the full 30 seconds, 100 requests are allowed again
349 1 : time::advance(time::Duration::from_millis(1)).await;
350 101 : for _ in 0..100 {
351 100 : assert!(limiter.check(endpoint, 1));
352 1 : }
353 1 : }
354 :
355 : #[tokio::test]
356 1 : async fn test_rate_limits_gc() {
357 1 : // fixed seeded random/hasher to ensure that the test is not flaky
358 1 : let rand = rand::rngs::StdRng::from_seed([1; 32]);
359 1 : let hasher = BuildHasherDefault::<FxHasher>::default();
360 1 :
361 1 : let limiter =
362 1 : BucketRateLimiter::new_with_rand_and_hasher(&RateBucketInfo::DEFAULT_SET, rand, hasher);
363 1000001 : for i in 0..1_000_000 {
364 1000000 : limiter.check(i, 1);
365 1000000 : }
366 1 : assert!(limiter.map.len() < 150_000);
367 1 : }
368 : }
|
