MDEV-26: Global transaction ID.

Fix error handling when record_gtid() fails to update the
mysql.rpl_slave_state table.

5 files changed, 108 insertions, 30 deletions

diff --git a/mysql-test/suite/rpl/r/rpl_gtid_errorhandling.result b/mysql-test/suite/rpl/r/rpl_gtid_errorhandling.result
index db6e2672a18..3ff6fe10e4a 100644
--- a/mysql-test/suite/rpl/r/rpl_gtid_errorhandling.result
+++ b/mysql-test/suite/rpl/r/rpl_gtid_errorhandling.result
@@ -7,21 +7,21 @@ ALTER TABLE mysql.rpl_slave_state CHANGE seq_no seq_no VARCHAR(20);
 START SLAVE;
 INSERT INTO t1 VALUES (1);
 CALL mtr.add_suppression("Slave: Failed to open mysql.rpl_slave_state");
-include/wait_for_slave_sql_error.inc [errno=1943]
+include/wait_for_slave_sql_error.inc [errno=1941]
 include/stop_slave.inc
 ALTER TABLE mysql.rpl_slave_state CHANGE seq_no seq_no BIGINT UNSIGNED NOT NULL;
 ALTER TABLE mysql.rpl_slave_state DROP PRIMARY KEY;
 ALTER TABLE mysql.rpl_slave_state ADD PRIMARY KEY (sub_id, domain_id);
 START SLAVE;
-include/wait_for_slave_sql_error.inc [errno=1943]
+include/wait_for_slave_sql_error.inc [errno=1941]
 include/stop_slave.inc
 ALTER TABLE mysql.rpl_slave_state DROP PRIMARY KEY;
 START SLAVE;
-include/wait_for_slave_sql_error.inc [errno=1943]
+include/wait_for_slave_sql_error.inc [errno=1941]
 include/stop_slave.inc
 ALTER TABLE mysql.rpl_slave_state ADD PRIMARY KEY (sub_id);
 START SLAVE;
-include/wait_for_slave_sql_error.inc [errno=1943]
+include/wait_for_slave_sql_error.inc [errno=1941]
 include/stop_slave.inc
 ALTER TABLE mysql.rpl_slave_state DROP PRIMARY KEY;
 ALTER TABLE mysql.rpl_slave_state ADD PRIMARY KEY (domain_id, sub_id);
@@ -85,5 +85,23 @@ START SLAVE;
 SET sql_log_bin=0;
 CALL mtr.add_suppression("The slave I/O thread stops because master does not support MariaDB global transaction id");
 SET sql_log_bin=1;
+*** Test error during record_gtid() (non-xid cases) ***
+include/stop_slave.inc
+CREATE TABLE t2 (a INT) ENGINE=MyISAM;
+INSERT INTO t2 VALUES (1);
+SET @old_dbug= @@global.DEBUG_DBUG;
+SET GLOBAL debug_dbug="+d,gtid_inject_record_gtid";
+START SLAVE;
+include/wait_for_slave_sql_error.inc [errno=1941]
+SET GLOBAL debug_dbug= @old_dbug;
+START SLAVE SQL_THREAD;
+SELECT * FROM t2;
+a
+1
+1
+SET sql_log_bin=0;
+CALL mtr.add_suppression("Slave: Could not update replication slave gtid state");
+SET sql_log_bin=1;
 DROP TABLE t1;
+DROP TABLE t2;
 include/rpl_end.inc
diff --git a/mysql-test/suite/rpl/t/rpl_gtid_errorhandling.test b/mysql-test/suite/rpl/t/rpl_gtid_errorhandling.test
index ada6fa22490..0efdcd18d35 100644
--- a/mysql-test/suite/rpl/t/rpl_gtid_errorhandling.test
+++ b/mysql-test/suite/rpl/t/rpl_gtid_errorhandling.test
@@ -16,7 +16,7 @@ INSERT INTO t1 VALUES (1);
 
 --connection slave
 CALL mtr.add_suppression("Slave: Failed to open mysql.rpl_slave_state");
---let $slave_sql_errno=1943
+--let $slave_sql_errno=1941
 --source include/wait_for_slave_sql_error.inc
 
 --source include/stop_slave.inc
@@ -24,19 +24,19 @@ ALTER TABLE mysql.rpl_slave_state CHANGE seq_no seq_no BIGINT UNSIGNED NOT NULL;
 ALTER TABLE mysql.rpl_slave_state DROP PRIMARY KEY;
 ALTER TABLE mysql.rpl_slave_state ADD PRIMARY KEY (sub_id, domain_id);
 START SLAVE;
---let $slave_sql_errno=1943
+--let $slave_sql_errno=1941
 --source include/wait_for_slave_sql_error.inc
 
 --source include/stop_slave.inc
 ALTER TABLE mysql.rpl_slave_state DROP PRIMARY KEY;
 START SLAVE;
---let $slave_sql_errno=1943
+--let $slave_sql_errno=1941
 --source include/wait_for_slave_sql_error.inc
 
 --source include/stop_slave.inc
 ALTER TABLE mysql.rpl_slave_state ADD PRIMARY KEY (sub_id);
 START SLAVE;
---let $slave_sql_errno=1943
+--let $slave_sql_errno=1941
 --source include/wait_for_slave_sql_error.inc
 
 --source include/stop_slave.inc
@@ -134,7 +134,37 @@ CALL mtr.add_suppression("The slave I/O thread stops because master does not sup
 SET sql_log_bin=1;
 
 
+--echo *** Test error during record_gtid() (non-xid cases) ***
+
+--connection slave
+--source include/stop_slave.inc
+
+--connection master
+CREATE TABLE t2 (a INT) ENGINE=MyISAM;
+INSERT INTO t2 VALUES (1);
+--save_master_pos
+
+--connection slave
+SET @old_dbug= @@global.DEBUG_DBUG;
+SET GLOBAL debug_dbug="+d,gtid_inject_record_gtid";
+
+START SLAVE;
+--let $slave_sql_errno= 1941
+--source include/wait_for_slave_sql_error.inc
+
+SET GLOBAL debug_dbug= @old_dbug;
+
+START SLAVE SQL_THREAD;
+--sync_with_master
+
+SELECT * FROM t2;
+SET sql_log_bin=0;
+CALL mtr.add_suppression("Slave: Could not update replication slave gtid state");
+SET sql_log_bin=1;
+
+
 --connection master
 DROP TABLE t1;
+DROP TABLE t2;
 
 --source include/rpl_end.inc
diff --git a/sql/log_event.cc b/sql/log_event.cc
index 1309907e91f..b083a6eb2a4 100644
--- a/sql/log_event.cc
+++ b/sql/log_event.cc
@@ -3814,26 +3814,6 @@ int Query_log_event::do_apply_event(Relay_log_info const *rli,
     */
     const_cast<Relay_log_info*>(rli)->inc_event_relay_log_pos();
     const_cast<Relay_log_info*>(rli)->clear_flag(Relay_log_info::IN_STMT);
-
-    /*
-      Record any GTID in the same transaction, so slave state is
-      transactionally consistent.
-    */
-    if ((sub_id= rli->gtid_sub_id))
-    {
-      /* Clear the GTID from the RLI so we don't accidentally reuse it. */
-      const_cast<Relay_log_info*>(rli)->gtid_sub_id= 0;
-
-      gtid= rli->current_gtid;
-      error= rpl_global_gtid_slave_state.record_gtid(thd, &gtid, sub_id, true);
-      if (error)
-      {
-        my_error(ER_CANNOT_UPDATE_GTID_STATE, MYF(0));
-        trans_rollback(thd);
-        sub_id= 0;
-        goto compare_errors;
-      }
-    }
   }
   else
   {
@@ -3958,6 +3938,30 @@ int Query_log_event::do_apply_event(Relay_log_info const *rli,
       else
         thd->variables.collation_database= thd->db_charset;
       
+      /*
+        Record any GTID in the same transaction, so slave state is
+        transactionally consistent.
+      */
+      if (strcmp("COMMIT", query) == 0 && (sub_id= rli->gtid_sub_id))
+      {
+        /* Clear the GTID from the RLI so we don't accidentally reuse it. */
+        const_cast<Relay_log_info*>(rli)->gtid_sub_id= 0;
+
+        gtid= rli->current_gtid;
+        if (rpl_global_gtid_slave_state.record_gtid(thd, &gtid, sub_id, true))
+        {
+          rli->report(ERROR_LEVEL, ER_CANNOT_UPDATE_GTID_STATE,
+                      "Error during COMMIT: failed to update GTID state in "
+                    "%s.%s: %d: %s",
+                      "mysql", rpl_gtid_slave_state_table_name.str,
+                      thd->stmt_da->sql_errno(), thd->stmt_da->message());
+          trans_rollback(thd);
+          sub_id= 0;
+          thd->is_slave_error= 1;
+          goto end;
+        }
+      }
+
       thd->table_map_for_update= (table_map)table_map_for_update;
       thd->set_invoker(&user, &host);
       /*
@@ -6842,7 +6846,13 @@ int Xid_log_event::do_apply_event(Relay_log_info const *rli)
     err= rpl_global_gtid_slave_state.record_gtid(thd, &gtid, sub_id, true);
     if (err)
     {
+      rli->report(ERROR_LEVEL, ER_CANNOT_UPDATE_GTID_STATE,
+                  "Error during XID COMMIT: failed to update GTID state in "
+                  "%s.%s: %d: %s",
+                  "mysql", rpl_gtid_slave_state_table_name.str,
+                  thd->stmt_da->sql_errno(), thd->stmt_da->message());
       trans_rollback(thd);
+      thd->is_slave_error= 1;
       return err;
     }
   }
diff --git a/sql/rpl_gtid.cc b/sql/rpl_gtid.cc
index 6a57f7cdb9c..7f5aa0b780b 100644
--- a/sql/rpl_gtid.cc
+++ b/sql/rpl_gtid.cc
@@ -29,7 +29,7 @@
 
 
 const LEX_STRING rpl_gtid_slave_state_table_name=
-  { STRING_WITH_LEN("rpl_slave_state") };
+  { C_STRING_WITH_LEN("rpl_slave_state") };
 
 
 void
@@ -298,6 +298,12 @@ rpl_slave_state::record_gtid(THD *thd, const rpl_gtid *gtid, uint64 sub_id,
 
   mysql_reset_thd_for_next_command(thd, 0);
 
+  DBUG_EXECUTE_IF("gtid_inject_record_gtid",
+                  {
+                    my_error(ER_CANNOT_UPDATE_GTID_STATE, MYF(0));
+                    return 1;
+                  } );
+
   tlist.init_one_table(STRING_WITH_LEN("mysql"),
                        rpl_gtid_slave_state_table_name.str,
                        rpl_gtid_slave_state_table_name.length,
@@ -328,6 +334,7 @@ rpl_slave_state::record_gtid(THD *thd, const rpl_gtid *gtid, uint64 sub_id,
   if ((elem= get_element(gtid->domain_id)) == NULL)
   {
     unlock();
+    my_error(ER_OUT_OF_RESOURCES, MYF(0));
     err= 1;
     goto end;
   }
diff --git a/sql/rpl_rli.cc b/sql/rpl_rli.cc
index 11249854c15..41bff96b5e5 100644
--- a/sql/rpl_rli.cc
+++ b/sql/rpl_rli.cc
@@ -1232,7 +1232,20 @@ void Relay_log_info::stmt_done(my_off_t event_master_log_pos,
   else
   {
     inc_group_relay_log_pos(event_master_log_pos);
-    rpl_global_gtid_slave_state.record_and_update_gtid(thd, this);
+    if (rpl_global_gtid_slave_state.record_and_update_gtid(thd, this))
+    {
+      report(WARNING_LEVEL, ER_CANNOT_UPDATE_GTID_STATE,
+             "Failed to update GTID state in %s.%s, slave state may become "
+             "inconsistent: %d: %s",
+             "mysql", rpl_gtid_slave_state_table_name.str,
+             thd->stmt_da->sql_errno(), thd->stmt_da->message());
+      /*
+        At this point we are not in a transaction (for example after DDL),
+        so we can not roll back. Anyway, normally updates to the slave
+        state table should not fail, and if they do, at least we made the
+        DBA aware of the problem in the error log.
+      */
+    }
     flush_relay_log_info(this);
     /*
       Note that Rotate_log_event::do_apply_event() does not call this