Uploaded image for project: 'MariaDB Server'
  1. MariaDB Server
  2. MDEV-22996

Hang on galera_toi_truncate test case

Details

    • Bug
    • Status: Closed (View Workflow)
    • Major
    • Resolution: Fixed
    • 10.4(EOL)
    • N/A
    • Galera, Tests
    • None

    Description

      galera.galera_toi_truncate 'innodb'      w4 [ fail ]  timeout after 900 seconds
              Test ended at 2020-06-24 09:40:02
       
      Test case timeout after 900 seconds
       
      == /home/jan/mysql/10.4/mysql-test/var/4/log/galera_toi_truncate.log == 
      connection node_2;
      connection node_1;
      connection node_1;
      CREATE TABLE ten (f1 INTEGER NOT NULL PRIMARY KEY) ENGINE=InnoDB;
      INSERT INTO ten VALUES (1),(2),(3),(4),(5),(6),(7),(8),(9),(10);
      CREATE TABLE t1 (f1 INTEGER) ENGINE=InnoDB;
      connection node_2;
      SET SESSION wsrep_retry_autocommit = 0;
      INSERT INTO t1 (f1) SELECT 1 FROM ten AS a1, ten AS a2, ten AS a3, ten AS a4, ten AS a5, ten AS a6, ten AS a7, ten AS a8;
      connection node_1;
      TRUNCATE TABLE t1;;
      connection node_2;
      

      Attachments

        Activity

          How to repeat (sometimes):

          • branch: 10.4 commit b80b52394d41a4c334642ae8b3af16f76f6fac57
          • galera library: 26.4.5
          • ./mtr --suite=wsrep,galera,galera_sr,galera_3nodes --force --parallel=4 --big-test
          jplindst Jan Lindström (Inactive) added a comment - How to repeat (sometimes): branch: 10.4 commit b80b52394d41a4c334642ae8b3af16f76f6fac57 galera library: 26.4.5 ./mtr --suite=wsrep,galera,galera_sr,galera_3nodes --force --parallel=4 --big-test

          The test is taking quite a lot of time also on 10.3. I aborted the execution after the server had consumed 14 CPU minutes. Some observations:

          • INSERT…SELECT in InnoDB will acquire explicit locks on the source table, making it very slow.
          • The Cartesian product on InnoDB tables in the test should better be replaced with something like INSERT INTO t1 SELECT 1 FROM seq_1_to_100000000.
          • Does the test really have to insert 10⁸ rows? Which scenario is it attempting to cover?
          • MDEV-13564 in MariaDB 10.2.19 could have radically affected the test scenario. Do we really need this test in its current form?
          marko Marko Mäkelä added a comment - The test is taking quite a lot of time also on 10.3. I aborted the execution after the server had consumed 14 CPU minutes. Some observations: INSERT…SELECT in InnoDB will acquire explicit locks on the source table, making it very slow. The Cartesian product on InnoDB tables in the test should better be replaced with something like INSERT INTO t1 SELECT 1 FROM seq_1_to_100000000 . Does the test really have to insert 10⁸ rows? Which scenario is it attempting to cover? MDEV-13564 in MariaDB 10.2.19 could have radically affected the test scenario. Do we really need this test in its current form?
          seppo Seppo Jaakola added a comment -

          a PR to fix the test will be submitted today

          seppo Seppo Jaakola added a comment - a PR to fix the test will be submitted today

          Found in disabled.def that it was disabled, also I'm adding the missing label.

          julien.fritsch Julien Fritsch added a comment - Found in disabled.def that it was disabled, also I'm adding the missing label.

          Found in disabled.def for ES that it was disabled, also I'm adding the missing label.

          julien.fritsch Julien Fritsch added a comment - Found in disabled.def for ES that it was disabled, also I'm adding the missing label.

          People

            jplindst Jan Lindström (Inactive)
            jplindst Jan Lindström (Inactive)
            Votes:
            0 Vote for this issue
            Watchers:
            4 Start watching this issue

            Dates

              Created:
              Updated:
              Resolved:

              Git Integration

                Error rendering 'com.xiplink.jira.git.jira_git_plugin:git-issue-webpanel'. Please contact your Jira administrators.