From 7c368e2c128f94e9dc7d250f1876c9b1c8899dcd Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tom=C3=A1s=20Gr=C3=BCner?=
 <47506558+MegaRedHand@users.noreply.github.com>
Date: Wed, 22 Jul 2026 12:26:16 -0300
Subject: [PATCH 1/2] fix(checkpoint-sync): retry failed attempts within
 startup budget

Checkpoint sync failed fast on any connection or timeout error: with a
single checkpoint URL, one 15s miss to a peer still booting crashed the
whole process (Hive reported "client did not start: terminated
unexpectedly"). Wrap the anchor fetch in a bounded retry that reattempts
a failed fetch a fixed number of times with a fixed backoff, staying
within Hive's client-startup budget. Per-attempt connect/read timeouts
and the internal AnchorPairingMismatch retry are unchanged.
---
 bin/ethlambda/src/checkpoint_sync.rs | 36 ++++++++++++++++++++++++++++
 bin/ethlambda/src/main.rs            |  2 +-
 2 files changed, 37 insertions(+), 1 deletion(-)

diff --git a/bin/ethlambda/src/checkpoint_sync.rs b/bin/ethlambda/src/checkpoint_sync.rs
index 9bed6372..970fe88a 100644
--- a/bin/ethlambda/src/checkpoint_sync.rs
+++ b/bin/ethlambda/src/checkpoint_sync.rs
@@ -27,6 +27,14 @@ const MAX_ANCHOR_FETCH_ATTEMPTS: u32 = 3;
 /// Delay between anchor fetch attempts.
 const ANCHOR_FETCH_RETRY_DELAY: Duration = Duration::from_secs(1);
 
+/// Fixed backoff between checkpoint-sync attempts.
+const CHECKPOINT_RETRY_BACKOFF: Duration = Duration::from_secs(5);
+
+/// Maximum checkpoint-sync attempts before the process gives up, giving a
+/// slow-to-boot peer time to become reachable. Attempts stay within Hive's
+/// client-startup budget when each one fails fast.
+const MAX_CHECKPOINT_ATTEMPTS: u32 = 5;
+
 #[derive(Debug, thiserror::Error)]
 pub enum CheckpointSyncError {
     #[error("HTTP request failed: {0}")]
@@ -346,6 +354,34 @@ pub async fn fetch_anchor_block_and_state(
     }
 }
 
+/// Fetch the finalized anchor, retrying any failure (e.g. the peer not yet
+/// reachable) for up to MAX_CHECKPOINT_ATTEMPTS attempts with a fixed backoff
+/// between them before giving up.
+pub async fn fetch_anchor_with_retry(
+    checkpoint_urls: &[String],
+    genesis_time: u64,
+    validators: &[Validator],
+) -> Result<(State, SignedBlock), CheckpointSyncError> {
+    let mut attempt: u32 = 1;
+    loop {
+        match fetch_anchor_block_and_state(checkpoint_urls, genesis_time, validators).await {
+            Ok(pair) => return Ok(pair),
+            Err(err) if attempt < MAX_CHECKPOINT_ATTEMPTS => {
+                warn!(
+                    attempt,
+                    max_attempts = MAX_CHECKPOINT_ATTEMPTS,
+                    %err,
+                    backoff_secs = CHECKPOINT_RETRY_BACKOFF.as_secs(),
+                    "Checkpoint sync attempt failed; retrying"
+                );
+                tokio::time::sleep(CHECKPOINT_RETRY_BACKOFF).await;
+                attempt += 1;
+            }
+            Err(err) => return Err(err),
+        }
+    }
+}
+
 #[cfg(test)]
 mod tests {
     use super::*;
diff --git a/bin/ethlambda/src/main.rs b/bin/ethlambda/src/main.rs
index b4d2828d..33e242e9 100644
--- a/bin/ethlambda/src/main.rs
+++ b/bin/ethlambda/src/main.rs
@@ -707,7 +707,7 @@ async fn fetch_initial_state(
 
     info!(?checkpoint_urls, "Starting checkpoint sync");
 
-    let (state, signed_block) = checkpoint_sync::fetch_anchor_block_and_state(
+    let (state, signed_block) = checkpoint_sync::fetch_anchor_with_retry(
         checkpoint_urls,
         genesis.genesis_time,
         &validators,

From d5bf94ede189947aff28ab69c768c5e068dbe4e8 Mon Sep 17 00:00:00 2001
From: =?UTF-8?q?Tom=C3=A1s=20Gr=C3=BCner?=
 <47506558+MegaRedHand@users.noreply.github.com>
Date: Wed, 22 Jul 2026 12:56:18 -0300
Subject: [PATCH 2/2] chore: simplify log

---
 bin/ethlambda/src/checkpoint_sync.rs | 8 +-------
 1 file changed, 1 insertion(+), 7 deletions(-)

diff --git a/bin/ethlambda/src/checkpoint_sync.rs b/bin/ethlambda/src/checkpoint_sync.rs
index 970fe88a..ff466ec1 100644
--- a/bin/ethlambda/src/checkpoint_sync.rs
+++ b/bin/ethlambda/src/checkpoint_sync.rs
@@ -367,13 +367,7 @@ pub async fn fetch_anchor_with_retry(
         match fetch_anchor_block_and_state(checkpoint_urls, genesis_time, validators).await {
             Ok(pair) => return Ok(pair),
             Err(err) if attempt < MAX_CHECKPOINT_ATTEMPTS => {
-                warn!(
-                    attempt,
-                    max_attempts = MAX_CHECKPOINT_ATTEMPTS,
-                    %err,
-                    backoff_secs = CHECKPOINT_RETRY_BACKOFF.as_secs(),
-                    "Checkpoint sync attempt failed; retrying"
-                );
+                warn!(attempt, %err, "Checkpoint sync attempt failed; retrying");
                 tokio::time::sleep(CHECKPOINT_RETRY_BACKOFF).await;
                 attempt += 1;
             }