From 07456c27f3ff9836c8684797c75889c2ea1f2125 Mon Sep 17 00:00:00 2001
From: atlas <atlas@hyperhive.local>
Date: Fri, 16 Oct 2026 14:11:19 +0200
Subject: [PATCH] start: log an unreadable applied-rev marker instead of
 silently treating it as stale

start_many mapped every read error on the marker to None via .ok(), so the
agent went down the rebuild path while the log blamed a stale rev. A missing
marker stays silent; any other error now logs the path and error. Control
flow is unchanged.
---
 hive-c0re/src/job_queue/power.rs | 32 +++++++++++++++++++++++++++++++-
 1 file changed, 31 insertions(+), 1 deletion(-)

diff --git a/hive-c0re/src/job_queue/power.rs b/hive-c0re/src/job_queue/power.rs
index 8e286896..fb47a602 100644
--- a/hive-c0re/src/job_queue/power.rs
+++ b/hive-c0re/src/job_queue/power.rs
@@ -291,7 +291,7 @@ pub async fn start_many(
     let mut targets = Vec::with_capacity(agents.len());
     for agent in agents {
         let running = lifecycle::is_running(agent).await;
-        let stored = std::fs::read_to_string(crate::paths::applied_rev_marker(agent)).ok();
+        let stored = read_rev_marker(&crate::paths::applied_rev_marker(agent));
         let stale = current
             .as_ref()
             .is_some_and(|rev| stored.as_deref() != Some(rev.as_str()));
@@ -305,6 +305,20 @@ pub async fn start_many(
     Ok(ids)
 }
 
+/// The rev `marker` records, or `None` when there is none. A missing file is
+/// the normal never-applied case; any other read error is logged, and the
+/// caller still treats the rev as stale.
+fn read_rev_marker(marker: &std::path::Path) -> Option<String> {
+    match std::fs::read_to_string(marker) {
+        Ok(rev) => Some(rev),
+        Err(e) if e.kind() == std::io::ErrorKind::NotFound => None,
+        Err(e) => {
+            tracing::warn!(path = %marker.display(), error = %e, "start: applied-rev marker unreadable; treating rev as stale");
+            None
+        }
+    }
+}
+
 /// Stop `agents` in a **single** DAG. A running agent gets
 /// `SetWanted(Off) → [Signal → Drain →](graceful) Reconcile`; a down agent
 /// skips the pointless quiesce but keeps the `Reconcile` as the race-up
@@ -354,3 +368,19 @@ pub fn pause_many(coord: &Arc<Coordinator>, agents: &[String]) -> anyhow::Result
     coord.emit_rebuild_queue_snapshot();
     Ok(ids)
 }
+
+#[cfg(test)]
+mod tests {
+    use super::read_rev_marker;
+
+    #[test]
+    fn a_rev_marker_reads_as_its_contents_and_none_when_absent_or_unreadable() {
+        let dir = tempfile::tempdir().unwrap();
+        let marker = dir.path().join("rev");
+        assert_eq!(read_rev_marker(&marker), None);
+        std::fs::write(&marker, "abc").unwrap();
+        assert_eq!(read_rev_marker(&marker).as_deref(), Some("abc"));
+        // A directory where the file belongs fails with a non-NotFound error.
+        assert_eq!(read_rev_marker(dir.path()), None);
+    }
+}