osd/scrub: restart snap trimming only after scrubbing is done

Snap trimming that was postponed as the target PG was scrubbing
must be restarted at scrub completion.
PR #38111 moved trimming restart to just before the scrub fully
terminated. The current PR fixes that.

Trimming is also restarted in those cases where scrub was
queued but aborted immediately.

Fixes: https://tracker.ceph.com/issues/52026

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

2 files changed, 9 insertions, 7 deletions

diff --git a/src/osd/PrimaryLogPG.cc b/src/osd/PrimaryLogPG.cc
index 043d1695f39..f4625a78cf3 100644
--- a/src/osd/PrimaryLogPG.cc
+++ b/src/osd/PrimaryLogPG.cc
@@ -4912,8 +4912,8 @@ void PrimaryLogPG::kick_snap_trim()
 
 void PrimaryLogPG::snap_trimmer_scrub_complete()
 {
-  if (is_primary() && is_active() && is_clean()) {
-    ceph_assert(!snap_trimq.empty());
+  if (is_primary() && is_active() && is_clean() && !snap_trimq.empty()) {
+    dout(10) << "scrub finished - requeuing snap_trimmer" << dendl;
     snap_trimmer_machine.process_event(ScrubComplete());
   }
 }
diff --git a/src/osd/scrubber/pg_scrubber.cc b/src/osd/scrubber/pg_scrubber.cc
index b7f007c7385..5dcc17b8138 100644
--- a/src/osd/scrubber/pg_scrubber.cc
+++ b/src/osd/scrubber/pg_scrubber.cc
@@ -189,6 +189,8 @@ void PgScrubber::initiate_regular_scrub(epoch_t epoch_queued)
     dout(10) << "scrubber event --<< StartScrub" << dendl;
   } else {
     clear_queued_or_active();
+    // and just in case snap trimming was blocked by the aborted scrub
+    m_pg->snap_trimmer_scrub_complete();
   }
 }
 
@@ -203,6 +205,8 @@ void PgScrubber::initiate_scrub_after_repair(epoch_t epoch_queued)
     dout(10) << "scrubber event --<< AfterRepairScrub" << dendl;
   } else {
     clear_queued_or_active();
+    // and just in case snap trimming was blocked by the aborted scrub
+    m_pg->snap_trimmer_scrub_complete();
   }
 }
 
@@ -1793,11 +1797,6 @@ void PgScrubber::scrub_finish()
       &t);
     int tr = m_osds->store->queue_transaction(m_pg->ch, std::move(t), nullptr);
     ceph_assert(tr == 0);
-
-    if (!m_pg->snap_trimq.empty()) {
-      dout(10) << "scrub finished, requeuing snap_trimmer" << dendl;
-      m_pg->snap_trimmer_scrub_complete();
-    }
   }
 
   if (has_error) {
@@ -1817,6 +1816,9 @@ void PgScrubber::scrub_finish()
   if (m_pg->is_active() && m_pg->is_primary()) {
     m_pg->recovery_state.share_pg_info();
   }
+
+  // we may have blocked the snap trimmer
+  m_pg->snap_trimmer_scrub_complete();
 }
 
 void PgScrubber::on_digest_updates()