Back off a failing feed exponentially, up to a day (#99)

A feed that failed was tried again on its usual schedule however long it had been failing:
gizmodo's 404, pelgrane's 403, daily-quests' 503 and toddstashwick's redirect loop every hour,
each a request to a site that had said no, a warning and scan time. A failing feed now waits as
long as it has been failing, from error_since to its last check, never less than its usual
interval and never more than a day: 1h, 1h, 2h, 4h, 8h, 16h, then daily on an hourly schedule.
No new column: error_since already marks the run's start and the first success clears it. A
forced refresh skips the due check, so it still tries at once. The feed list's next check
follows the backoff.

Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>

CHANGELOG.md

@@ -23,6 +23,7 @@ and this project adheres to [Semantic Versioning](https://semver.org/spec/v2.0.0
 
 ### Changed
 
+- A feed that keeps failing is checked less and less often, waiting as long as it has been failing, up to once a day; it goes back to its schedule as soon as it works. Refreshing it still checks it at once.
 - A scan's trace shows the time a feed spends on its artwork and in the database after the fetch.
 - The JSON log carries each line's `trace_id` and `span_id`, logs each scan event once instead of
   twice, names a failure's kind in `error.type` (and its HTTP status in

docs/configuration.md

@@ -45,7 +45,9 @@ media_types       = ["audio", "video"]
 database as described above; `download_dir`, `socket` and `organize` stay in config.toml.
 
 * **`schedule`** — how often feeds are re-checked. A feed's own `<ttl>` still wins when it asks to
-  be polled *less* often, and a per-feed `schedule` overrides both. Admin-only from the UI.
+  be polled *less* often, and a per-feed `schedule` overrides both. A feed that keeps failing
+  waits as long as it has been failing before the next try, up to a day, and is back on schedule
+  after its first success. Admin-only from the UI.
 * **`organize`** — `feed` files downloads under the feed's folder; `date` under `YYYY-MM-DD`.
 * **`max_total_gb`** — the reaper deletes to get back under this, oldest first, keeping a 50 MB
   pad. Kept items are never deleted, and a file only counts as read once every subscriber has

src/db.rs

@@ -415,6 +415,8 @@ pub struct HttpState {
     pub last_modified: Option<String>,
     pub last_checked: Option<i64>,
     pub ttl_mins: Option<u64>,
+    /// When the feed's current run of failures began, for backing off (`due_after`).
+    pub error_since: Option<i64>,
 }
 
 impl Db {
@@ -427,6 +429,7 @@ impl Db {
                 last_modified: f.last_modified,
                 last_checked: f.last_checked,
                 ttl_mins: f.ttl_mins.map(|t| t.max(0) as u64),
+                error_since: f.error_since,
             })
             .unwrap_or_default())
     }

src/main.rs

@@ -981,7 +981,7 @@ async fn fetch(ctx: &Arc<Ctx>, only: Option<&str>, force: bool, scope: &[String]
         }
 
         if !force && let Some(last) = state.last_checked {
-            let due = last + due_after(&cfg, feed_cfg, state.ttl_mins) as i64;
+            let due = last + due_after(&cfg, feed_cfg, state.ttl_mins, state.error_since, last) as i64;
             if due > db::now() {
                 ctx.out.emit(Event::FeedSkip {
                     feed: id.clone(),
@@ -1165,13 +1165,34 @@ async fn retire_stranded(ctx: &Ctx) -> Result<usize> {
 ///
 /// A per-feed schedule is an explicit instruction and wins outright. Without one, the
 /// global schedule applies, but the feed's own <ttl> raises it when the publisher asks to
-/// be polled less often.
-pub fn due_after(cfg: &config::Config, feed: &config::Feed, ttl_mins: Option<u64>) -> u64 {
+/// be polled less often. A feed that is failing backs off (`backoff`), from `error_since`, when
+/// its run of failures began, to `last_checked`, when it last failed.
+pub fn due_after(
+    cfg: &config::Config,
+    feed: &config::Feed,
+    ttl_mins: Option<u64>,
+    error_since: Option<i64>,
+    last_checked: i64,
+) -> u64 {
     let mins = match feed.schedule.as_deref().and_then(config::parse_interval) {
         Some(explicit) => explicit,
         None => ttl_mins.unwrap_or(0).max(cfg.general.interval()),
     };
-    mins * 60
+    let failing_for = error_since.map(|since| (last_checked - since).max(0) as u64);
+    backoff(mins * 60, failing_for)
+}
+
+/// A failing feed waits as long as it has been failing, so the wait doubles with each failure
+/// (1h, 1h, 2h, 4h, ... on an hourly schedule), never less than its usual interval and, past that,
+/// never more than a day. Retried hourly, a feed dead for good cost a request, a warning and scan
+/// time every hour (#99). The first success clears error_since, and with it the backoff; a
+/// refresh someone asks for is not held back by it.
+fn backoff(usual: u64, failing_for: Option<u64>) -> u64 {
+    const CEILING: u64 = 86_400;
+    match failing_for {
+        Some(f) => usual.max(f.min(CEILING)),
+        None => usual,
+    }
 }
 
 #[derive(Default)]
@@ -1836,6 +1857,22 @@ fn duration(secs: u64) -> String {
 mod tests {
     use super::*;
 
+    #[test]
+    fn a_failing_feed_backs_off_doubling_up_to_a_day() {
+        let hour = 3600;
+        assert_eq!(backoff(hour, None), hour);
+        // Failing since 0: each check waits as long as the failure has lasted so far.
+        let (mut at, mut waits) = (0, vec![]);
+        for _ in 0..8 {
+            let w = backoff(hour, Some(at));
+            waits.push(w / hour);
+            at += w;
+        }
+        assert_eq!(waits, [1, 1, 2, 4, 8, 16, 24, 24]);
+        // A weekly schedule is longer than the ceiling and stays as it is.
+        assert_eq!(backoff(7 * 86_400, Some(30 * 86_400)), 7 * 86_400);
+    }
+
     #[tokio::test]
     async fn the_first_start_moves_the_configuration_in_and_trims_the_file() {
         let dir = std::env::temp_dir().join(format!("ipx-assemble-{}", std::process::id()));

src/web.rs

@@ -790,11 +790,11 @@ async fn feeds(
                 .schedule
                 .as_deref()
                 .and_then(crate::config::parse_interval),
-            every_mins: crate::due_after(&cfg, feed, ttl_mins) / 60,
+            every_mins: crate::due_after(&cfg, feed, ttl_mins, None, 0) / 60,
             last_checked: s.last_checked,
             next_check: s
                 .last_checked
-                .map(|t| t + crate::due_after(&cfg, feed, ttl_mins) as i64),
+                .map(|t| t + crate::due_after(&cfg, feed, ttl_mins, s.error_since, t) as i64),
             failing: s
                 .error_since
                 .filter(|since| crate::db::now() - since >= FLAG_AFTER_SECS)