Merge MDEV-8147 into 10.0

mysql-test/suite/rpl/r/rpl_parallel_partition.result

+include/master-slave.inc
+[connection master]
+SET @old_parallel_threads=@@GLOBAL.slave_parallel_threads;
+include/stop_slave.inc
+SET @old_dbug= @@GLOBAL.debug_dbug;
+SET GLOBAL debug_dbug="+d,inject_wakeup_subsequent_commits_sleep";
+SET GLOBAL slave_parallel_threads=8;
+*** MDEV-8147: Assertion `m_lock_type == 2' failed in handler::ha_close() during parallel replication ***
+CREATE TABLE E (
+pk INTEGER AUTO_INCREMENT,
+col_int_nokey INTEGER /*! NULL */,
+col_int_key INTEGER /*! NULL */,
+col_date_key DATE /*! NULL */,
+col_date_nokey DATE /*! NULL */,
+col_time_key TIME /*! NULL */,
+col_time_nokey TIME /*! NULL */,
+col_datetime_key DATETIME /*! NULL */,
+col_datetime_nokey DATETIME /*! NULL */,
+col_varchar_key VARCHAR(1) /*! NULL */,
+col_varchar_nokey VARCHAR(1) /*! NULL */,
+PRIMARY KEY (pk),
+KEY (col_int_key),
+KEY (col_date_key),
+KEY (col_time_key),
+KEY (col_datetime_key),
+KEY (col_varchar_key, col_int_key)
+)  ENGINE=InnoDB;
+ALTER TABLE `E` PARTITION BY KEY() PARTITIONS 5;
+ALTER TABLE `E` REMOVE PARTITIONING;
+CREATE TABLE t1 (a INT PRIMARY KEY);
+include/start_slave.inc
+include/stop_slave.inc
+SET GLOBAL debug_dbug=@old_dbug;
+SET GLOBAL slave_parallel_threads=0;
+SET GLOBAL slave_parallel_threads=8;
+include/start_slave.inc
+include/stop_slave.inc
+SET GLOBAL slave_parallel_threads=@old_parallel_threads;
+include/start_slave.inc
+DROP TABLE `E`;
+DROP TABLE t1;
+include/rpl_end.inc

mysql-test/suite/rpl/t/rpl_parallel_partition.test

+--source include/have_partition.inc
+--source include/have_innodb.inc
+--source include/have_debug.inc
+--source include/master-slave.inc
+
+--connection server_2
+SET @old_parallel_threads=@@GLOBAL.slave_parallel_threads;
+--source include/stop_slave.inc
+SET @old_dbug= @@GLOBAL.debug_dbug;
+SET GLOBAL debug_dbug="+d,inject_wakeup_subsequent_commits_sleep";
+SET GLOBAL slave_parallel_threads=8;
+
+--echo *** MDEV-8147: Assertion `m_lock_type == 2' failed in handler::ha_close() during parallel replication ***
+--connection server_1
+CREATE TABLE E (
+			pk INTEGER AUTO_INCREMENT,
+			col_int_nokey INTEGER /*! NULL */,
+			col_int_key INTEGER /*! NULL */,
+
+			col_date_key DATE /*! NULL */,
+			col_date_nokey DATE /*! NULL */,
+
+			col_time_key TIME /*! NULL */,
+			col_time_nokey TIME /*! NULL */,
+
+			col_datetime_key DATETIME /*! NULL */,
+			col_datetime_nokey DATETIME /*! NULL */,
+
+			col_varchar_key VARCHAR(1) /*! NULL */,
+			col_varchar_nokey VARCHAR(1) /*! NULL */,
+
+			PRIMARY KEY (pk),
+			KEY (col_int_key),
+			KEY (col_date_key),
+			KEY (col_time_key),
+			KEY (col_datetime_key),
+			KEY (col_varchar_key, col_int_key)
+		)  ENGINE=InnoDB;
+
+ALTER TABLE `E` PARTITION BY KEY() PARTITIONS 5;
+ALTER TABLE `E` REMOVE PARTITIONING;
+--write_file $MYSQLTEST_VARDIR/tmp/mysqld.1.expect
+wait
+EOF
+--shutdown_server 30
+--source include/wait_until_disconnected.inc
+--connection default
+--source include/wait_until_disconnected.inc
+--append_file $MYSQLTEST_VARDIR/tmp/mysqld.1.expect
+restart:
+EOF
+--enable_reconnect
+--source include/wait_until_connected_again.inc
+--connection server_1
+--enable_reconnect
+--source include/wait_until_connected_again.inc
+CREATE TABLE t1 (a INT PRIMARY KEY);
+--save_master_pos
+
+--connection server_2
+--source include/start_slave.inc
+--sync_with_master
+
+# Re-spawn worker threads to clear dbug injection.
+--source include/stop_slave.inc
+SET GLOBAL debug_dbug=@old_dbug;
+SET GLOBAL slave_parallel_threads=0;
+SET GLOBAL slave_parallel_threads=8;
+--source include/start_slave.inc
+
+
+# Clean up.
+--connection server_2
+--source include/stop_slave.inc
+SET GLOBAL slave_parallel_threads=@old_parallel_threads;
+--source include/start_slave.inc
+
+--connection server_1
+DROP TABLE `E`;
+DROP TABLE t1;
+--source include/rpl_end.inc

sql/mysqld.cc

@@ -9513,6 +9513,7 @@ PSI_stage_info stage_waiting_for_work_from_sql_thread= { 0, "Waiting for work fr
 PSI_stage_info stage_waiting_for_prior_transaction_to_commit= { 0, "Waiting for prior transaction to commit", 0};
 PSI_stage_info stage_waiting_for_prior_transaction_to_start_commit= { 0, "Waiting for prior transaction to start commit before starting next transaction", 0};
 PSI_stage_info stage_waiting_for_room_in_worker_thread= { 0, "Waiting for room in worker thread event queue", 0};
+PSI_stage_info stage_waiting_for_workers_idle= { 0, "Waiting for worker threads to be idle", 0};
 PSI_stage_info stage_master_gtid_wait_primary= { 0, "Waiting in MASTER_GTID_WAIT() (primary waiter)", 0};
 PSI_stage_info stage_master_gtid_wait= { 0, "Waiting in MASTER_GTID_WAIT()", 0};
 PSI_stage_info stage_gtid_wait_other_connection= { 0, "Waiting for other master connection to process GTID received on multiple master connections", 0};

sql/mysqld.h

@@ -453,6 +453,7 @@ extern PSI_stage_info stage_waiting_for_work_from_sql_thread;
 extern PSI_stage_info stage_waiting_for_prior_transaction_to_commit;
 extern PSI_stage_info stage_waiting_for_prior_transaction_to_start_commit;
 extern PSI_stage_info stage_waiting_for_room_in_worker_thread;
+extern PSI_stage_info stage_waiting_for_workers_idle;
 extern PSI_stage_info stage_master_gtid_wait_primary;
 extern PSI_stage_info stage_master_gtid_wait;
 extern PSI_stage_info stage_gtid_wait_other_connection;

sql/rpl_parallel.cc

@@ -168,6 +168,8 @@ finish_event_group(rpl_parallel_thread *rpt, uint64 sub_id,
       done and also no longer need waiting for.
     */
     entry->last_committed_sub_id= sub_id;
+    if (entry->need_sub_id_signal)
+      mysql_cond_broadcast(&entry->COND_parallel_entry);
 
     /* Now free any GCOs in which all transactions have committed. */
     group_commit_orderer *tmp_gco= rgi->gco;
@@ -1894,26 +1896,29 @@ rpl_parallel::wait_for_workers_idle(THD *thd)
   max_i= domain_hash.records;
   for (i= 0; i < max_i; ++i)
   {
-    bool active;
-    wait_for_commit my_orderer;
+    PSI_stage_info old_stage;
     struct rpl_parallel_entry *e;
+    int err= 0;
 
     e= (struct rpl_parallel_entry *)my_hash_element(&domain_hash, i);
     mysql_mutex_lock(&e->LOCK_parallel_entry);
-    if ((active= (e->current_sub_id > e->last_committed_sub_id)))
-    {
-      wait_for_commit *waitee= &e->current_group_info->commit_orderer;
-      my_orderer.register_wait_for_prior_commit(waitee);
-      thd->wait_for_commit_ptr= &my_orderer;
-    }
-    mysql_mutex_unlock(&e->LOCK_parallel_entry);
-    if (active)
+    e->need_sub_id_signal= true;
+    thd->ENTER_COND(&e->COND_parallel_entry, &e->LOCK_parallel_entry,
+                    &stage_waiting_for_workers_idle, &old_stage);
+    while (e->current_sub_id > e->last_committed_sub_id)
     {
-      int err= my_orderer.wait_for_prior_commit(thd);
-      thd->wait_for_commit_ptr= NULL;
-      if (err)
-        return err;
+      if (thd->check_killed())
+      {
+        thd->send_kill_message();
+        err= 1;
+        break;
+      }
+      mysql_cond_wait(&e->COND_parallel_entry, &e->LOCK_parallel_entry);
     }
+    e->need_sub_id_signal= false;
+    thd->EXIT_COND(&old_stage);
+    if (err)
+      return err;
   }
   return 0;
 }

sql/rpl_parallel.h

@@ -227,6 +227,12 @@ struct rpl_parallel_entry {
     waiting for event groups to complete.
   */
   bool force_abort;
+  /*
+    Set in wait_for_workers_idle() to show that it is waiting, so that
+    finish_event_group knows to signal it when last_committed_sub_id is
+    increased.
+  */
+  bool need_sub_id_signal;
   /*
    At STOP SLAVE (force_abort=true), we do not want to process all events in
    the queue (which could unnecessarily delay stop, if a lot of events happen

sql/sql_class.cc

@@ -6623,6 +6623,7 @@ wait_for_commit::wakeup_subsequent_commits2(int wakeup_error)
     a mutex), so no extra explicit barrier is needed here.
   */
   wakeup_subsequent_commits_running= false;
+  DBUG_EXECUTE_IF("inject_wakeup_subsequent_commits_sleep", my_sleep(21000););
 }