From 9d50e7bc26d6ab164f3e1534f36a08d57a2be02a Mon Sep 17 00:00:00 2001 From: CPerezz Date: Sun, 8 Feb 2026 08:51:10 +0100 Subject: [PATCH] eth/protocols/snap: restore peers to idle pool on request revert All five revert*Request functions (account, bytecode, storage, trienode heal, bytecode heal) remove the request from the tracked set but never restore the peer to its corresponding idle pool. When a request times out and no response arrives, the peer is permanently lost from the idle pool, preventing new work from being assigned to it. In normal operation this bug is masked by pivot movement (which resets idle pools via new Sync() cycles every ~15 minutes) and peer churn (reconnections re-add peers via Register()). However in scenarios with long-running sync cycles and few peers, all peers can eventually leak out of the idle pools, stalling sync entirely. Fix: after deleting from the request map, restore the peer to its idle pool if it is still registered (guards against the peer-drop path where Unregister already removed the peer). This mirrors the pattern used in all five On* response handlers. --- eth/protocols/snap/sync.go | 30 +++++++++++++++++++++++++----- 1 file changed, 25 insertions(+), 5 deletions(-) diff --git a/eth/protocols/snap/sync.go b/eth/protocols/snap/sync.go index cf4e494645..2645233e93 100644 --- a/eth/protocols/snap/sync.go +++ b/eth/protocols/snap/sync.go @@ -1699,9 +1699,13 @@ func (s *Syncer) revertAccountRequest(req *accountRequest) { } close(req.stale) - // Remove the request from the tracked set + // Remove the request from the tracked set and restore the peer to the + // idle pool so it can be reassigned work (skip if peer already left). s.lock.Lock() delete(s.accountReqs, req.id) + if _, ok := s.peers[req.peer]; ok { + s.accountIdlers[req.peer] = struct{}{} + } s.lock.Unlock() // If there's a timeout timer still running, abort it and mark the account @@ -1740,9 +1744,13 @@ func (s *Syncer) revertBytecodeRequest(req *bytecodeRequest) { } close(req.stale) - // Remove the request from the tracked set + // Remove the request from the tracked set and restore the peer to the + // idle pool so it can be reassigned work (skip if peer already left). s.lock.Lock() delete(s.bytecodeReqs, req.id) + if _, ok := s.peers[req.peer]; ok { + s.bytecodeIdlers[req.peer] = struct{}{} + } s.lock.Unlock() // If there's a timeout timer still running, abort it and mark the code @@ -1781,9 +1789,13 @@ func (s *Syncer) revertStorageRequest(req *storageRequest) { } close(req.stale) - // Remove the request from the tracked set + // Remove the request from the tracked set and restore the peer to the + // idle pool so it can be reassigned work (skip if peer already left). s.lock.Lock() delete(s.storageReqs, req.id) + if _, ok := s.peers[req.peer]; ok { + s.storageIdlers[req.peer] = struct{}{} + } s.lock.Unlock() // If there's a timeout timer still running, abort it and mark the storage @@ -1826,9 +1838,13 @@ func (s *Syncer) revertTrienodeHealRequest(req *trienodeHealRequest) { } close(req.stale) - // Remove the request from the tracked set + // Remove the request from the tracked set and restore the peer to the + // idle pool so it can be reassigned work (skip if peer already left). s.lock.Lock() delete(s.trienodeHealReqs, req.id) + if _, ok := s.peers[req.peer]; ok { + s.trienodeHealIdlers[req.peer] = struct{}{} + } s.lock.Unlock() // If there's a timeout timer still running, abort it and mark the trie node @@ -1867,9 +1883,13 @@ func (s *Syncer) revertBytecodeHealRequest(req *bytecodeHealRequest) { } close(req.stale) - // Remove the request from the tracked set + // Remove the request from the tracked set and restore the peer to the + // idle pool so it can be reassigned work (skip if peer already left). s.lock.Lock() delete(s.bytecodeHealReqs, req.id) + if _, ok := s.peers[req.peer]; ok { + s.bytecodeHealIdlers[req.peer] = struct{}{} + } s.lock.Unlock() // If there's a timeout timer still running, abort it and mark the code