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