Files
maarcadetweet/crates/at-firehose/src/consumer.rs
T
tomdeboneandClaude Opus 5 b58cb75cfe fix(at-firehose): Jetstream-Filter wirkte nie — Collections in die URL
Der Consumer verband sich auf die nackte URL und schickte danach
`{"type":"options","wantedCollections":[…]}` als Textframe. Jetstream
ignoriert das, und zwar stillschweigend: Filter sind Query-Parameter, und
der einzige nachrichtenbasierte Weg (`options_update`) verlangt, dass die
Verbindung mit `requireHello=true` geöffnet wurde.

Jede Instanz, die glaubte, sechs Collections zu abonnieren, hat also den
kompletten öffentlichen Firehose gezogen. Gemessen gegen
jetstream1.us-east: 3119 Events in 8 s ungefiltert, 520 für eine einzelne
Collection, 12 für die beiden, die dieses Projekt wirklich braucht.

Konkrete Folgen: die Dev-Datenbank ist unbemerkt auf 3,3 Mio. Posts
gewachsen, und auf der Produktionsinstanz musste die AppView abgeschaltet
und aus dem Autostart genommen werden, weil sie die Platte vollzuschreiben
drohte — dort stand `JETSTREAM_COLLECTIONS=app.twi.post` korrekt in der
.env und wurde einfach nicht beachtet.

Was der Fix NICHT löst, und das steht auch so im Code: die Collections, die
ein Bluesky-artiges Produkt normalerweise will (post/like/repost/follow),
sind ~97 % des Volumens. Richtig zu filtern ist notwendig, nicht
hinreichend.

Nebenbei: upload_blob_rejects_oversized fiel etwa jeden dritten Lauf um. Der
Server bricht die Verbindung ab, sobald das Body-Limit reißt, also sieht der
Client je nach Timing die 413 oder einen Reset beim Schreiben. Beides
beweist, dass der Upload abgelehnt wurde; der Test akzeptiert jetzt beides.

Co-Authored-By: Claude Opus 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_013HC9HLrUU1LNwkzp8nkDLX
2026-09-10 20:30:20 +02:00

196 lines
7.2 KiB
Rust

use anyhow::Result;
use futures::StreamExt;
use std::sync::Arc;
use std::sync::atomic::{AtomicBool, Ordering};
use std::time::Duration;
use tokio_tungstenite::tungstenite::Message;
use tracing::{error, info, warn};
use crate::event::JetstreamEvent;
pub struct JetstreamConsumer {
pub url: String,
pub collections: Vec<String>,
pub max_backoff_secs: u64,
/// Optional handle the consumer toggles on every connect/disconnect.
/// Useful for health endpoints that want a live "are we connected?"
/// signal without polling.
pub connected: Option<Arc<AtomicBool>>,
/// Optional cursor (`time_us`) to resume from. 0 means "no cursor / start
/// fresh" which is Jetstream's default. Set via `with_cursor(...)`.
pub cursor_us: i64,
}
impl JetstreamConsumer {
pub fn new(url: impl Into<String>, collections: Vec<String>) -> Self {
Self {
url: url.into(),
collections,
max_backoff_secs: 30,
connected: None,
cursor_us: 0,
}
}
/// The URL actually dialled: base URL plus `wantedCollections` and
/// `cursor` as query parameters.
///
/// This used to connect to the bare URL and then send
/// `{"type": "options", "wantedCollections": [...]}` as a text frame.
/// Jetstream ignores that, and silently: filters are query parameters,
/// and the only message-based path (`options_update`) requires the
/// connection to have been opened with `requireHello=true`. So every
/// deployment that thought it was subscribing to six collections was in
/// fact taking the entire public firehose — measured against
/// jetstream1.us-east: 3119 events in 8 s unfiltered versus 520 for a
/// single collection. On the dev database that quietly grew to 3.3 M
/// posts; on the production instance the AppView had to be switched off
/// to stop it filling the disk.
///
/// Note what fixing this does *not* solve: the collections this project
/// wants (`app.bsky.feed.post` / `like` / `repost` / `graph.follow`) are
/// ~97 % of the firehose by volume. Correct filtering is necessary, not
/// sufficient — an instance that does not want the whole public network
/// in its index wants `wantedDids`, or no Jetstream at all.
pub fn subscribe_url(&self) -> String {
let mut url = self.url.trim_end_matches('&').to_string();
let mut sep = if url.contains('?') { '&' } else { '?' };
for c in &self.collections {
url.push(sep);
url.push_str("wantedCollections=");
url.push_str(&urlencode(c));
sep = '&';
}
if self.cursor_us > 0 {
url.push(sep);
url.push_str("cursor=");
url.push_str(&self.cursor_us.to_string());
}
url
}
/// Build a consumer that shares a connection-state flag with the caller.
pub fn with_connected_flag(mut self, flag: Arc<AtomicBool>) -> Self {
self.connected = Some(flag);
self
}
/// Build a consumer with an explicit reconnect backoff ceiling (seconds).
pub fn with_max_backoff_secs(mut self, max: u64) -> Self {
self.max_backoff_secs = max.max(1);
self
}
/// Build a consumer that starts from a previously-persisted cursor
/// (microseconds since epoch). Setting this avoids the post-restart
/// gap where Jetstream's default backfill window might miss events.
pub fn with_cursor(mut self, cursor_us: i64) -> Self {
self.cursor_us = cursor_us;
self
}
pub async fn run<F, Fut>(&self, mut on_event: F) -> Result<()>
where
F: FnMut(JetstreamEvent) -> Fut + Send + 'static,
Fut: std::future::Future<Output = Result<()>> + Send,
{
let mut backoff_secs: u64 = 1;
loop {
let r = self.connect_and_consume(&mut on_event).await;
// Any non-error return (clean disconnect, error) means we lost
// the connection; flip the flag if we own one and back off.
if let Some(flag) = &self.connected {
flag.store(false, Ordering::Relaxed);
}
match r {
Ok(()) => warn!("jetstream stream ended, reconnecting"),
Err(e) => error!("jetstream error: {e:#}"),
}
warn!("reconnecting in {backoff_secs}s");
tokio::time::sleep(Duration::from_secs(backoff_secs)).await;
backoff_secs = (backoff_secs * 2).min(self.max_backoff_secs);
}
}
async fn connect_and_consume<F, Fut>(&self, on_event: &mut F) -> Result<()>
where
F: FnMut(JetstreamEvent) -> Fut + Send,
Fut: std::future::Future<Output = Result<()>> + Send,
{
let url = self.subscribe_url();
let (mut ws, _) = tokio_tungstenite::connect_async(&url).await?;
info!("connected to jetstream: {url}");
if let Some(flag) = &self.connected {
flag.store(true, Ordering::Relaxed);
}
while let Some(msg) = ws.next().await {
let msg = msg?;
if let Message::Text(text) = msg {
if let Ok(ev) = serde_json::from_str::<JetstreamEvent>(&text) {
on_event(ev).await?;
}
}
}
Ok(())
}
}
/// Percent-encode everything outside the unreserved set. Collection NSIDs are
/// dots and letters today, but a `wantedDids` value carries `:` — encoding
/// unconditionally keeps this correct if the caller passes one.
fn urlencode(s: &str) -> String {
s.chars()
.map(|c| match c {
'A'..='Z' | 'a'..='z' | '0'..='9' | '-' | '_' | '.' | '~' => c.to_string(),
other => {
let mut buf = [0u8; 4];
other
.encode_utf8(&mut buf)
.bytes()
.map(|b| format!("%{b:02X}"))
.collect()
}
})
.collect()
}
#[cfg(test)]
mod url_tests {
use super::*;
#[test]
fn collections_go_into_the_query_string() {
let c = JetstreamConsumer::new(
"wss://jetstream1.us-east.bsky.network/subscribe",
vec!["app.twi.post".into(), "app.bsky.feed.like".into()],
);
assert_eq!(
c.subscribe_url(),
"wss://jetstream1.us-east.bsky.network/subscribe\
?wantedCollections=app.twi.post&wantedCollections=app.bsky.feed.like"
.replace(' ', "")
);
}
#[test]
fn cursor_is_appended_and_respects_an_existing_query() {
let mut c = JetstreamConsumer::new("wss://host/subscribe?compress=false", vec![]);
c.cursor_us = 1234;
assert_eq!(c.subscribe_url(), "wss://host/subscribe?compress=false&cursor=1234");
}
#[test]
fn no_filters_leaves_the_url_alone() {
let c = JetstreamConsumer::new("wss://host/subscribe", vec![]);
assert_eq!(c.subscribe_url(), "wss://host/subscribe");
}
/// A DID contains `:`, which has to survive as `%3A` in a query value.
#[test]
fn values_are_percent_encoded() {
assert_eq!(urlencode("did:plc:abc"), "did%3Aplc%3Aabc");
assert_eq!(urlencode("app.bsky.feed.post"), "app.bsky.feed.post");
}
}