osd/SnapMapper::update_snaps() to handle a missing OBJ_ record

by falling back to add_oid() instead of silently creating an
inconsistent state (OBJ_ without matching SNA_ entries). This
was observed on replicas that had recently recovered objects:
the snap mapper entries created during recovery were not visible
to a subsequent snap-trim repop's update_snaps() call, leaving
the clone with no snap mapper entries. Scrub would then detect
and report the inconsistency as an error.

Promote snap mapper remove_oid/clear_snaps logging to dout(10)
and add apply_op_stats tracing to aid diagnosis of any remaining
stat or snap mapper drift.

Signed-off-by: Ronen Friedman <rfriedma@redhat.com>

diff --git a/src/osd/SnapMapper.cc b/src/osd/SnapMapper.cc
index df2da64b3e499830d3d4dfc1c4d49d2544ce53a8..0faf77a5b3908f745df1c7ec2bb49a89f9fb9e11 100644 (file)
--- a/src/osd/SnapMapper.cc
+++ b/src/osd/SnapMapper.cc
@@ -522,7 +522,7 @@ void SnapMapper::clear_snaps(
   const hobject_t &oid,
   MapCacher::Transaction<std::string, ceph::buffer::list> *t)
 {
-  dout(20) << __func__ << " " << oid << dendl;
+  dout(10) << __func__ << " " << oid << dendl;
   ceph_assert(check(oid));
   set<string> to_remove;
   to_remove.insert(to_object_key(oid));
@@ -589,6 +589,15 @@ int SnapMapper::update_snaps(
   // Tolerate missing keys but not disk errors
   if (r < 0 && r != -ENOENT)
     return r;
+  if (r == -ENOENT) {
+    // Recently recovered replicas may observe missing snap-mapper state.
+    // Avoid creating an inconsistent state (OBJ_ without matching SNA_ entries) that
+    // would be detected by scrub later on - instead rebuild the mapping from scratch.
+    dout(10) << fmt::format("{}: {} no existing snap mapping, creating for {}",
+              __func__, oid, new_snaps) << dendl;
+    add_oid(oid, new_snaps, t);
+    return 0;
+  }
   if (old_snaps_check)
     ceph_assert(out.snaps == *old_snaps_check);
 
@@ -770,7 +779,7 @@ int SnapMapper::remove_oid(
   const hobject_t &oid,
   MapCacher::Transaction<std::string, ceph::buffer::list> *t)
 {
-  dout(20) << *this << __func__ << " " << oid << dendl;
+  dout(10) << *this << __func__ << " " << oid << dendl;
   ceph_assert(check(oid));
   return _remove_oid(oid, t);
 }