Uploaded image for project: 'MariaDB Server'
  1. MariaDB Server
  2. MDEV-27131

Add option for SQL thread to kill any blocking local activity that blocks replicated DDL

Details

    • New Feature
    • Status: Closed (View Workflow)
    • Major
    • Resolution: Fixed
    • N/A
    • Replication
    • None

    Description

      DDL executed on primary may be blocked by any long running SELECT on replica, thus breaking the "high availability" of this setup. It seems there is no way even to force limited wait time for DDL on replica.

      Please, add server variable to force priority for SQL thread in such cases, similar to what happens in Galera.

      UPDATE:
      from MDEV-29569, which is the same request, but has one additional information:
      "Proper logging options (similar to those we have in Galera) should also be added, to help understand the reason of error local connection to slave may get."

      Attachments

        Issue Links

          is duplicated by

          Task - A task that needs to be done. MDEV-29569 Add option for SQL thread to terminate all transactions and statements that block it

          • Major - Major loss of function.
          • Closed

          New Feature - A new feature of the product, which has yet to be developed. MDEV-34857 Implement --slave-abort-blocking-timeout

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

          New Feature - A new feature of the product, which has yet to be developed. MDEV-36302 Add option for SQL thread to kill any blocking DML

          • Major - Major loss of function.
          • Open

          Activity

            Ascending order - Click to sort in descending order
            Elkin Andrei Elkin added a comment - - edited

            ralf.gebhardt@mariadb.com, the plan has been to implement a similar to an existing thd_rpl_deadlock_check() extension to MDL deadlock mechanism.
            The replicated DDL that is MDL requestor would effectively kill a local connection in its way.
            I made some experimenting to prove the concept.

            Elkin Andrei Elkin added a comment - - edited ralf.gebhardt@mariadb.com , the plan has been to implement a similar to an existing thd_rpl_deadlock_check() extension to MDL deadlock mechanism. The replicated DDL that is MDL requestor would effectively kill a local connection in its way. I made some experimenting to prove the concept.
            monty Michael Widenius added a comment -

            This is fixed as part of MDEV-34857.
            It was my fault that a duplicate Jira entry was created.

            monty Michael Widenius added a comment - This is fixed as part of MDEV-34857 . It was my fault that a duplicate Jira entry was created.
            knielsen Kristian Nielsen added a comment -

            This functionality is available from MDEV-34857 --slave-abort-blocking-timeout, which is now pushed to 11.7.
            So maybe this one can be closed.

            knielsen Kristian Nielsen added a comment - This functionality is available from MDEV-34857 --slave-abort-blocking-timeout, which is now pushed to 11.7. So maybe this one can be closed.

            People

              knielsen Kristian Nielsen
              valerii Valerii Kravchuk
              Votes:
              1 Vote for this issue
              Watchers:
              8 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.