this repo has no description
lewis.moe
1use crate::circuit_breaker::CircuitBreaker;
2use crate::sync::firehose::SequencedEvent;
3use reqwest::Client;
4use std::sync::Arc;
5use std::sync::atomic::{AtomicU64, Ordering};
6use std::time::Duration;
7use tokio::sync::{broadcast, watch};
8use tracing::{debug, error, info, warn};
9
10const NOTIFY_THRESHOLD_SECS: u64 = 20 * 60;
11
12pub struct Crawlers {
13 hostname: String,
14 crawler_urls: Vec<String>,
15 http_client: Client,
16 last_notified: AtomicU64,
17 circuit_breaker: Option<Arc<CircuitBreaker>>,
18}
19
20impl Crawlers {
21 pub fn new(hostname: String, crawler_urls: Vec<String>) -> Self {
22 Self {
23 hostname,
24 crawler_urls,
25 http_client: Client::builder()
26 .timeout(Duration::from_secs(30))
27 .build()
28 .unwrap_or_default(),
29 last_notified: AtomicU64::new(0),
30 circuit_breaker: None,
31 }
32 }
33
34 pub fn with_circuit_breaker(mut self, circuit_breaker: Arc<CircuitBreaker>) -> Self {
35 self.circuit_breaker = Some(circuit_breaker);
36 self
37 }
38
39 pub fn from_env() -> Option<Self> {
40 let hostname = std::env::var("PDS_HOSTNAME").ok()?;
41
42 let crawler_urls: Vec<String> = std::env::var("CRAWLERS")
43 .unwrap_or_default()
44 .split(',')
45 .filter(|s| !s.is_empty())
46 .map(|s| s.trim().to_string())
47 .collect();
48
49 if crawler_urls.is_empty() {
50 return None;
51 }
52
53 Some(Self::new(hostname, crawler_urls))
54 }
55
56 fn should_notify(&self) -> bool {
57 let now = std::time::SystemTime::now()
58 .duration_since(std::time::UNIX_EPOCH)
59 .unwrap_or_default()
60 .as_secs();
61
62 let last = self.last_notified.load(Ordering::Relaxed);
63 now - last >= NOTIFY_THRESHOLD_SECS
64 }
65
66 fn mark_notified(&self) {
67 let now = std::time::SystemTime::now()
68 .duration_since(std::time::UNIX_EPOCH)
69 .unwrap_or_default()
70 .as_secs();
71
72 self.last_notified.store(now, Ordering::Relaxed);
73 }
74
75 pub async fn notify_of_update(&self) {
76 if !self.should_notify() {
77 debug!("Skipping crawler notification due to debounce");
78 return;
79 }
80
81 if let Some(cb) = &self.circuit_breaker
82 && !cb.can_execute().await {
83 debug!("Skipping crawler notification due to circuit breaker open");
84 return;
85 }
86
87 self.mark_notified();
88 let circuit_breaker = self.circuit_breaker.clone();
89
90 for crawler_url in &self.crawler_urls {
91 let url = format!(
92 "{}/xrpc/com.atproto.sync.requestCrawl",
93 crawler_url.trim_end_matches('/')
94 );
95 let hostname = self.hostname.clone();
96 let client = self.http_client.clone();
97 let cb = circuit_breaker.clone();
98
99 tokio::spawn(async move {
100 match client
101 .post(&url)
102 .json(&serde_json::json!({ "hostname": hostname }))
103 .send()
104 .await
105 {
106 Ok(response) => {
107 if response.status().is_success() {
108 debug!(crawler = %url, "Successfully notified crawler");
109 if let Some(cb) = cb {
110 cb.record_success().await;
111 }
112 } else {
113 let status = response.status();
114 let body = response.text().await.unwrap_or_default();
115 warn!(
116 crawler = %url,
117 status = %status,
118 body = %body,
119 hostname = %hostname,
120 "Crawler notification returned non-success status"
121 );
122 if let Some(cb) = cb {
123 cb.record_failure().await;
124 }
125 }
126 }
127 Err(e) => {
128 warn!(crawler = %url, error = %e, "Failed to notify crawler");
129 if let Some(cb) = cb {
130 cb.record_failure().await;
131 }
132 }
133 }
134 });
135 }
136 }
137}
138
139pub async fn start_crawlers_service(
140 crawlers: Arc<Crawlers>,
141 mut firehose_rx: broadcast::Receiver<SequencedEvent>,
142 mut shutdown: watch::Receiver<bool>,
143) {
144 info!(
145 hostname = %crawlers.hostname,
146 crawler_count = crawlers.crawler_urls.len(),
147 crawlers = ?crawlers.crawler_urls,
148 "Starting crawlers notification service"
149 );
150
151 loop {
152 tokio::select! {
153 result = firehose_rx.recv() => {
154 match result {
155 Ok(event) => {
156 if event.event_type == "commit" {
157 crawlers.notify_of_update().await;
158 }
159 }
160 Err(broadcast::error::RecvError::Lagged(n)) => {
161 warn!(skipped = n, "Crawlers service lagged behind firehose");
162 crawlers.notify_of_update().await;
163 }
164 Err(broadcast::error::RecvError::Closed) => {
165 error!("Firehose channel closed, stopping crawlers service");
166 break;
167 }
168 }
169 }
170 _ = shutdown.changed() => {
171 if *shutdown.borrow() {
172 info!("Crawlers service shutting down");
173 break;
174 }
175 }
176 }
177 }
178}