Uploaded image for project: 'MariaDB Connector/J'
  1. MariaDB Connector/J
  2. CONJ-676

Behavior of MariaDB 10.3.11 batch update processing changed from non-atomic insert to atomic insert

    XMLWordPrintable

Details

    • Bug
    • Status: Closed (View Workflow)
    • Critical
    • Resolution: Won't Fix
    • 2.2.3
    • N/A
    • batch
    • Linux 3.10.0-862.el7.x86_64
      MariaDB server 10.3.11
      Connector/J 2.2.3

    Description

      When we switched from MariaDB 10.1.16 server to Mariadb 10.3.11, the behavior of MariaDB batch update processing changed from non-atomic insert to atomic insert. This is an upward compatibility issue for us.
      In the repro2 I attached, we attempt to insert a block of 3 rows, one of which has a duplicate primary key value.
      With MaraiaDB server 10.1.16, the non-duplicate rows of a block insert would be inserted up to the first error. With MariaDB server 10.3.11, no rows are inserted and we don't know which row of the block caused the error, so we cannot attempt to insert all non-error rows. This looks like a bug. Is there a setting to restore the original behavior?

      Attachments

        1. repro2.class
          6 kB
        2. repro2.java
          8 kB
        3. repro2.log.mariadb10.1.16
          2 kB
        4. repro2.log.mariadb10.3.11
          2 kB
        5. setup
          0.1 kB

        Activity

          People

            Unassigned Unassigned
            Gofman Yekaterina
            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.