Uploaded image for project: 'MariaDB MaxScale'
  1. MariaDB MaxScale
  2. MXS-4586

transaction_replay_max_size default is 1GiB instead of 1MiB

    XMLWordPrintable

Details

    • Bug
    • Status: Closed (View Workflow)
    • Major
    • Resolution: Fixed
    • 2.5.25, 6.2.1, 22.08.5, 23.02.0, 23.02.1
    • 2.5.26, 6.4.7, 22.08.6, 23.02.2
    • readwritesplit
    • Running in a VM - Ubuntu 20.04.6 LTS, 2 CPUs, 4GB RAM (8GB for production)
    • MXS-SPRINT-181, MXS-SPRINT-182

    Description

      The memory usage of MaxScale keeps growing and eventually swap is used and then MaxScale gets busted one night (or day).

      Issue observed on our pre-prod MaxScale - version 23.02.1. The same observed in prod (6.2.1). Currently, holding prod upgrade.

      Stack trace has been logged (attached). Next step is using jemalloc to profile the heap as Markus suggested.

      Attachments

        1. maxscale.cnf
          2 kB
        2. maxscale.log
          32 kB
        3. screenshot-1.png
          screenshot-1.png
          429 kB
        4. screenshot-2.png
          screenshot-2.png
          332 kB
        5. maxscale-1.log
          80 kB
        6. heap-report.pdf
          15 kB
        7. result.txt
          247 kB
        8. Screenshot 2023-05-12 at 2.56.08 pm.png
          Screenshot 2023-05-12 at 2.56.08 pm.png
          129 kB
        9. maxscale.log.zip
          1.39 MB

        Issue Links

          Activity

            People

              nantti Niclas Antti
              eoin Eoin Kim
              Votes:
              1 Vote for this issue
              Watchers:
              3 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.