os/bluestore: fix _extend_log seq advance

when extending the log, the sequence was left on a bad state because it would first create a transaction to update with the current seq number but leave the "real" transaction with the same sequence number which should be `extend_log_transaction.seq + 1`.

Signed-off-by: Pere Diaz Bou <pdiabou@redhat.com>
(cherry picked from commit 63f0a0df14c9d8e68be61e374438bc75cef45a1f)

Fixes: https://tracker.ceph.com/issues/69764

diff --git a/src/os/bluestore/BlueFS.cc b/src/os/bluestore/BlueFS.cc
index 666a38b6b29efc156af5c2a7b76f54bcff227a6e..4c8a38b4b6d25e98d3b31509f6920642b0e8164e 100644 (file)
--- a/src/os/bluestore/BlueFS.cc
+++ b/src/os/bluestore/BlueFS.cc
@@ -3133,12 +3133,13 @@ void BlueFS::_extend_log(uint64_t amount) {
   _pad_bl(bl, super.block_size);
   log.writer->append(bl);
   ceph_assert(allocated_before_extension >= log.writer->get_effective_write_pos());
-  log.t.seq = log.seq_live;
 
   // before sync_core we advance the seq
   {
     std::unique_lock<ceph::mutex> l(dirty.lock);
-    _log_advance_seq();
+    dirty.seq_live++;
+    log.seq_live++;
+    log.t.seq++;
   }
 }