Uploaded image for project: 'MariaDB Server'
  1. MariaDB Server
  2. MDEV-6180

Error 1590 is not autoskippable

        Details

        • Type: Bug
        • Status: Closed
        • Priority: Minor
        • Resolution: Fixed
        • Affects Version/s: 10.0.10
        • Fix Version/s: 10.0.13
        • Component/s: None
        • Labels:
          None
        • Environment:
          standard Linux, using Debian packages from mariadb

          Description

          Because of a full disk, my master logged a large heap of "LOST" incident messages. I tried to auto-skip them, but that did not work.

          MariaDB [(none)]> show variables like '%slave_skip%';
+------------------------+-------+
| Variable_name          | Value |
+------------------------+-------+
| slave_skip_errors      | 1590  |
| sql_slave_skip_counter | 0     |
+------------------------+-------+
2 rows in set (0.00 sec)
MariaDB [(none)]> show slave status\G
            Slave_SQL_Running: No
                   Last_Errno: 1590
                   Last_Error: The incident LOST_EVENTS occured on the master. Message: error writing to the binary log

            Gliffy Diagrams

              Attachments

                Activity

                Ascending order - Click to sort in descending order
                Hide
                Permalink
                elenst Elena Stepanova added a comment -

                Test case to reproduce it without full disk
                (the binary log is attached, it was produced by MySQL 5.6 which creates LOST_EVENTS upon this scenario):

                On MySQL 5.6, it fails if run in the usual way, and passes if run with --mysqld=--slave-skip-errors=1590. On MariaDB 10.0, it fails either way.

                Put the attached log into mysql-test/std_data.
                Run with --mysqld=--slave-skip-errors=1590

                --source include/master-slave.inc

--connection slave
STOP SLAVE;
--source include/wait_for_slave_to_stop.inc
RESET SLAVE;

--connection master
RESET MASTER;
--let $datadir = `SELECT @@datadir`
--remove_file $datadir/master-bin.000001
--copy_file $MYSQL_TEST_DIR/std_data/mdev6180-bin.000001 $datadir/master-bin.000001
--save_master_pos

--connection slave
START SLAVE;
--let $slave_timeout = 5
--source include/wait_for_slave_to_start.inc
--sync_with_master
                Show
                elenst Elena Stepanova added a comment - Test case to reproduce it without full disk (the binary log is attached, it was produced by MySQL 5.6 which creates LOST_EVENTS upon this scenario): On MySQL 5.6, it fails if run in the usual way, and passes if run with --mysqld=--slave-skip-errors=1590 . On MariaDB 10.0, it fails either way. Put the attached log into mysql-test/std_data. Run with --mysqld=--slave-skip-errors=1590 --source include/master-slave.inc --connection slave STOP SLAVE; --source include/wait_for_slave_to_stop.inc RESET SLAVE; --connection master RESET MASTER; --let $datadir = `SELECT @@datadir` --remove_file $datadir/master-bin.000001 --copy_file $MYSQL_TEST_DIR/std_data/mdev6180-bin.000001 $datadir/master-bin.000001 --save_master_pos --connection slave START SLAVE; --let $slave_timeout = 5 --source include/wait_for_slave_to_start.inc --sync_with_master
                Hide
                Permalink
                knielsen Kristian Nielsen added a comment -

                Here is a MySQL 5.6 patch that fixes this:

                revision-id: frazer@mysql.com-20110314170916-ypgin17otj3ucx95
                parent: andrei.elkin@oracle.com-20110314135544-xez4ef4e0rynbmy9
                committer: Frazer Clement <frazer@mysql.com>
                branch nick: mysql-trunk
                timestamp: Mon 2011-03-14 17:09:16 +0000
                message:
                Bug#11799671 NOT POSSIBLE TO SKIP INCIDENT ERRORS

                do_apply_event() for the Incident_log_event is enhanced to check whether
                the event should be skipped.

                Testcase rpl_skip_incident is added.

                Show
                knielsen Kristian Nielsen added a comment - Here is a MySQL 5.6 patch that fixes this: revision-id: frazer@mysql.com-20110314170916-ypgin17otj3ucx95 parent: andrei.elkin@oracle.com-20110314135544-xez4ef4e0rynbmy9 committer: Frazer Clement <frazer@mysql.com> branch nick: mysql-trunk timestamp: Mon 2011-03-14 17:09:16 +0000 message: Bug#11799671 NOT POSSIBLE TO SKIP INCIDENT ERRORS do_apply_event() for the Incident_log_event is enhanced to check whether the event should be skipped. Testcase rpl_skip_incident is added.
                Hide
                Permalink
                knielsen Kristian Nielsen added a comment -

                I merged the MySQL 5.6 fix for Bug#11799671 into MariaDB 10.0.13

                Show
                knielsen Kristian Nielsen added a comment - I merged the MySQL 5.6 fix for Bug#11799671 into MariaDB 10.0.13

                  People

                  • Assignee:
                    knielsen Kristian Nielsen
                    Reporter:
                    smurf Matthias Urlichs
                  • Votes:
                    0 Vote for this issue
                    Watchers:
                    2 Start watching this issue

                    Dates

                    • Created:
                      Updated:
                      Resolved: