Uploaded image for project: 'MariaDB Server'
  1. MariaDB Server
  2. MDEV-25622

Insanely large incorrect values in performance_schema.file_summary_by_instance

    XMLWordPrintable

Details

    • Bug
    • Status: Open (View Workflow)
    • Major
    • Resolution: Unresolved
    • 10.3.27, 10.5.10
    • None
    • Performance Schema
    • None
    • Linux CentOS 8

    Description

      Sometimes file_summary_by_instance table of performance_schema reports insanely large values.

      For example, when I restart the server (to reset all stats) everything seems to work fine, but suddenly after a few minutes or hours I get something like io usage of more than a few month or even years!

      I attached a query and it's result that just a few minutes after a restart on a development machine, it shows that 3 internal tables and a wordpress table are using more than 4 months of io usage! As COUNT_STAR column shows, I just refreshed a just-installed WordPress page for a few times and suddenly this happened. Also I saw this issue on a Stack Overflow question a few days ago too, but I can't find a link to it.

      This is a handy important feature of MariaDB for both finding bottlenecks and also finding abusive users in multi-user environments. It's so sad that such a good feature is not only underrated, but also is buggy.

      I checked this with default setting on both 10.3.27 and 10.5.10 on CentOS, both versions have same problem.
      Please let me know if there is a need for more information.

      Attachments

        Activity

          People

            Unassigned Unassigned
            masoud Masoud
            Votes:
            0 Vote for this issue
            Watchers:
            2 Start watching this issue

            Dates

              Created:
              Updated:

              Git Integration

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