Uploaded image for project: 'MariaDB Server'
  1. MariaDB Server
  2. MDEV-18648

slave_parallel_mode= optimistic default in 10.5

Details

    Description

      To change CONSERVATIVE @@global.slave_parallel_mode default to OPTIMISTIC in 10.5.

      Attachments

        Issue Links

          is duplicated by

          Task - A task that needs to be done. MDEV-10244 Change default to use optimistic mode of in-order parallel replication

          • Major - Major loss of function.
          • Closed
          relates to

          Bug - A problem which impairs or prevents the functions of the product. MDEV-20605 Awaken transaction can miss inserted by other transaction records due to wrong persistent cursor restoration

          • Critical - Crashes, loss of data, severe memory leak.
          • Closed

          Activity

            Ascending order - Click to sort in descending order
            GeoffMontee Geoff Montee (Inactive) added a comment -

            Elkin mentioned to me that optimistic and aggressive mode tend to be more reliable and less likely to hit bugs when binlog_format is set to ROW. This is because these modes rely on internal rollbacks and retries in order to properly work, and these internal rollbacks and retries are more likely to have unexpected side effects when binlog_format is set to STATEMENT or MIXED.

            If we change the default value for slave_parallel_mode to optimistic in 10.5, then should we also change the default value of binlog_format to ROW?

            GeoffMontee Geoff Montee (Inactive) added a comment - Elkin mentioned to me that optimistic and aggressive mode tend to be more reliable and less likely to hit bugs when binlog_format is set to ROW . This is because these modes rely on internal rollbacks and retries in order to properly work, and these internal rollbacks and retries are more likely to have unexpected side effects when binlog_format is set to STATEMENT or MIXED. If we change the default value for slave_parallel_mode to optimistic in 10.5, then should we also change the default value of binlog_format to ROW ?
            Elkin Andrei Elkin added a comment -

            GeoffMontee: that's a good point, and did not weigh my MIXED preference against the "looming" OPTIMISTIC.
            Let us discuss your proposal with architecture, QA and replication teams.

            Elkin Andrei Elkin added a comment - GeoffMontee : that's a good point, and did not weigh my MIXED preference against the "looming" OPTIMISTIC. Let us discuss your proposal with architecture, QA and replication teams.
            sujatha.sivakumar Sujatha Sivakumar (Inactive) added a comment -

            Hello Andrei,

            Can you please review the MDEV-18648 changes.

            https://github.com/MariaDB/server/commit/cde9170709cec4d64ecf8c4c162be851223206fd

            http://buildbot.askmonty.org/buildbot/grid?category=main&branch=bb-10.5-sujatha

            Thank you

            sujatha.sivakumar Sujatha Sivakumar (Inactive) added a comment - Hello Andrei, Can you please review the MDEV-18648 changes. https://github.com/MariaDB/server/commit/cde9170709cec4d64ecf8c4c162be851223206fd http://buildbot.askmonty.org/buildbot/grid?category=main&branch=bb-10.5-sujatha Thank you
            Elkin Andrei Elkin added a comment -

            GeoffMontee: just for the records, bugs like MDEV-20605 were meant. ROW format load is free from it.

            Elkin Andrei Elkin added a comment - GeoffMontee : just for the records, bugs like MDEV-20605 were meant. ROW format load is free from it.

            People

              alice Alice Sherepa
              Elkin Andrei Elkin
              Votes:
              0 Vote for this issue
              Watchers:
              6 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.