MDEV-17801: Galera test failure on galera_var_reject_queries

Problem was that controlling connection i.e. connection that
executed the query SET GLOBAL wsrep_reject_queries = ALL_KILL;
was also killed but server would try to send result from that
query to controlling connection resulting a assertion
mysqld: /home/jan/mysql/10.2-sst/include/mysql/psi/mysql_socket.h:738: inline_mysql_socket_send: Assertion `mysql_socket.fd != -1' failed.
as socket was closed when controlling connection was closed.

wsrep_close_client_connections()
	Do not close controlling connection and instead of
	wsrep_close_thread() we do now soft kill by THD::awake

wsrep_reject_queries_update()
	Call wsrep_close_client_connections using current thd.

5 files changed, 28 insertions, 9 deletions

diff --git a/mysql-test/suite/galera/r/galera_var_reject_queries.result b/mysql-test/suite/galera/r/galera_var_reject_queries.result
index 5958e6ae981..1c932eb5509 100644
--- a/mysql-test/suite/galera/r/galera_var_reject_queries.result
+++ b/mysql-test/suite/galera/r/galera_var_reject_queries.result
@@ -5,7 +5,6 @@ SET GLOBAL wsrep_reject_queries = ALL;
 SELECT * FROM t1;
 ERROR 08S01: WSREP has not yet prepared node for application use
 SET GLOBAL wsrep_reject_queries = ALL_KILL;
-ERROR HY000: Lost connection to MySQL server during query
 SELECT * FROM t1;
 Got one of the listed errors
 SELECT * FROM t1;
diff --git a/mysql-test/suite/galera/t/galera_var_reject_queries.test b/mysql-test/suite/galera/t/galera_var_reject_queries.test
index 6859855c35f..8b80c04e3be 100644
--- a/mysql-test/suite/galera/t/galera_var_reject_queries.test
+++ b/mysql-test/suite/galera/t/galera_var_reject_queries.test
@@ -18,8 +18,11 @@ SET GLOBAL wsrep_reject_queries = ALL;
 --error ER_UNKNOWN_COM_ERROR
 SELECT * FROM t1;
 
-# Lost connection
---error 2013
+#
+# Original behavior was lost connection,
+# but since 10.1, we allow controlling connection to remain alive
+#
+--error 0,2013
 SET GLOBAL wsrep_reject_queries = ALL_KILL;
 
 --connection node_1a
diff --git a/sql/wsrep_mysqld.cc b/sql/wsrep_mysqld.cc
index c3f5dd1ce97..47aea6d3824 100644
--- a/sql/wsrep_mysqld.cc
+++ b/sql/wsrep_mysqld.cc
@@ -2229,7 +2229,7 @@ int wsrep_wait_committing_connections_close(int wait_time)
 }
 
 
-void wsrep_close_client_connections(my_bool wait_to_end)
+void wsrep_close_client_connections(my_bool wait_to_end, THD *except_caller_thd)
 {
   /*
     First signal all threads that it's time to die
@@ -2251,6 +2251,12 @@ void wsrep_close_client_connections(my_bool wait_to_end)
     if (!is_client_connection(tmp))
       continue;
 
+    if (tmp == except_caller_thd)
+    {
+      DBUG_ASSERT(is_client_connection(tmp));
+      continue;
+    }
+
     if (is_replaying_connection(tmp))
     {
       tmp->set_killed(KILL_CONNECTION);
@@ -2262,7 +2268,16 @@ void wsrep_close_client_connections(my_bool wait_to_end)
       continue;
 
     WSREP_DEBUG("closing connection %ld", tmp->thread_id);
-    wsrep_close_thread(tmp);
+
+    /*
+      instead of wsrep_close_thread() we do now  soft kill by THD::awake
+     */
+    mysql_mutex_lock(&tmp->LOCK_thd_data);
+
+    tmp->awake(KILL_CONNECTION);
+
+    mysql_mutex_unlock(&tmp->LOCK_thd_data);
+
   }
   mysql_mutex_unlock(&LOCK_thread_count);
 
@@ -2280,7 +2295,8 @@ void wsrep_close_client_connections(my_bool wait_to_end)
 #ifndef __bsdi__				// Bug in BSDI kernel
     if (is_client_connection(tmp) &&
         !abort_replicated(tmp)    &&
-	!is_replaying_connection(tmp))
+        !is_replaying_connection(tmp) &&
+        tmp != except_caller_thd)
     {
       WSREP_INFO("killing local connection: %ld",tmp->thread_id);
       close_connection(tmp,0);
diff --git a/sql/wsrep_mysqld.h b/sql/wsrep_mysqld.h
index 43f0d53393d..2814f4013d8 100644
--- a/sql/wsrep_mysqld.h
+++ b/sql/wsrep_mysqld.h
@@ -161,7 +161,6 @@ extern "C" query_id_t wsrep_thd_query_id(THD *thd);
 extern "C" query_id_t wsrep_thd_wsrep_last_query_id(THD *thd);
 extern "C" void wsrep_thd_set_wsrep_last_query_id(THD *thd, query_id_t id);
 
-extern void wsrep_close_client_connections(my_bool wait_to_end);
 extern int  wsrep_wait_committing_connections_close(int wait_time);
 extern void wsrep_close_applier(THD *thd);
 extern void wsrep_wait_appliers_close(THD *thd);
@@ -314,7 +313,8 @@ void thd_binlog_trx_reset(THD * thd);
 typedef void (*wsrep_thd_processor_fun)(THD *);
 pthread_handler_t start_wsrep_THD(void *arg);
 int wsrep_wait_committing_connections_close(int wait_time);
-void wsrep_close_client_connections(my_bool wait_to_end);
+extern void wsrep_close_client_connections(my_bool wait_to_end,
+                                           THD *except_caller_thd = NULL);
 void wsrep_close_applier(THD *thd);
 void wsrep_close_applier_threads(int count);
 void wsrep_wait_appliers_close(THD *thd);
diff --git a/sql/wsrep_var.cc b/sql/wsrep_var.cc
index 718676025f9..3959746156a 100644
--- a/sql/wsrep_var.cc
+++ b/sql/wsrep_var.cc
@@ -398,7 +398,8 @@ bool wsrep_reject_queries_update(sys_var *self, THD* thd, enum_var_type type)
             WSREP_INFO("Rejecting client queries due to manual setting");
             break;
         case WSREP_REJECT_ALL_KILL:
-            wsrep_close_client_connections(FALSE);
+            /* close all client connections, but this one */
+            wsrep_close_client_connections(FALSE, thd);
             WSREP_INFO("Rejecting client queries and killing connections due to manual setting");
             break;
         default: