MDEV-27760 event may non stop replicate in circular semisync setup

MDEV-21117 had to relax own events acceptance condition for a case
when a former semisync master server recovers after crash as the
semisync slave. That however admitted a possibility for endless event
"orbiting" in the non-strict slave gtid mode of semisync circular
setup.

The same server-id event termination is restored now for
the non-strict gtid mode to follow regular rules (that is it's ignored
unless @@global.replicate_same_server_id allows it in).

To address MDEV-21117 recovery agenda,
in the strict gtid mode and the transaction's gtid ordered strictly
greater than the current slave gtid state, the same server-id
transaction is accepted.

The gtid strict mode is safe to accept transactions even if
the slave state were not set correct by the user, e.g
at the former master.
An added test shows a typical out-of-order error at execution so
no data corruption is guaranteed in such a case.

1 files changed, 15 insertions, 9 deletions

diff --git a/sql/slave.cc b/sql/slave.cc
index c0eef02ca7a..a6578be199a 100644
--- a/sql/slave.cc
+++ b/sql/slave.cc
@@ -6186,13 +6186,13 @@ static int queue_event(Master_info* mi, const uchar *buf, ulong event_len)
   bool is_rows_event= false;
   /*
     The flag has replicate_same_server_id semantics and is raised to accept
-    a same-server-id event on the semisync slave, for both the gtid and legacy
-    connection modes.
-    Such events can appear as result of this server recovery so the event
-    was created there and replicated elsewhere right before the crash. At recovery
-    it could be evicted from the server's binlog.
-  */
-  bool do_accept_own_server_id= false;
+    a same-server-id event group by the gtid strict mode semisync slave.
+    Own server-id events can appear as result of this server crash-recovery:
+    the transaction was created on this server then being master, got replicated
+    elsewhere right before the crash before commit;
+    finally at recovery the transaction gets evicted from the server's binlog.
+ */
+  bool do_accept_own_server_id;
   /*
     FD_q must have been prepared for the first R_a event
     inside get_master_version_and_clock()
@@ -6281,6 +6281,8 @@ static int queue_event(Master_info* mi, const uchar *buf, ulong event_len)
                     dbug_rows_event_count = 0;
                   };);
 #endif
+  s_id= uint4korr(buf + SERVER_ID_OFFSET);
+
   mysql_mutex_lock(&mi->data_lock);
 
   switch (buf[EVENT_TYPE_OFFSET]) {
@@ -6722,6 +6724,7 @@ static int queue_event(Master_info* mi, const uchar *buf, ulong event_len)
 
     ++mi->events_queued_since_last_gtid;
     inc_pos= event_len;
+
   }
   break;
   /*
@@ -6864,6 +6867,10 @@ static int queue_event(Master_info* mi, const uchar *buf, ulong event_len)
     break;
   }
 
+  do_accept_own_server_id= (s_id == global_system_variables.server_id
+    && rpl_semi_sync_slave_enabled && opt_gtid_strict_mode
+    && mi->using_gtid != Master_info::USE_GTID_NO);
+
   /*
     Integrity of Rows- event group check.
     A sequence of Rows- events must end with STMT_END_F flagged one.
@@ -6909,7 +6916,6 @@ static int queue_event(Master_info* mi, const uchar *buf, ulong event_len)
   */
 
   mysql_mutex_lock(log_lock);
-  s_id= uint4korr(buf + SERVER_ID_OFFSET);
   /*
     Write the event to the relay log, unless we reconnected in the middle
     of an event group and now need to skip the initial part of the group that
@@ -6955,7 +6961,7 @@ static int queue_event(Master_info* mi, const uchar *buf, ulong event_len)
   else
   if ((s_id == global_system_variables.server_id &&
        !(mi->rli.replicate_same_server_id ||
-         (do_accept_own_server_id= rpl_semi_sync_slave_enabled))) ||
+         do_accept_own_server_id)) ||
       event_that_should_be_ignored(buf) ||
       /*
         the following conjunction deals with IGNORE_SERVER_IDS, if set