Uploaded image for project: 'MariaDB MaxScale'
  1. MariaDB MaxScale
  2. MXS-827

ReadWriteSplit only keeps used connection alive, query crashes after unused connection times out

    XMLWordPrintable

Details

    • New Feature
    • Status: Closed (View Workflow)
    • Major
    • Resolution: Fixed
    • 1.4.3
    • 2.2.0
    • readwritesplit
    • None
    • centos 7
      wait_timeout=60

    Description

      Problem

      after 60s maxscale sends "MySQL server has gone away" to the client

      Setup

      • wait_timeout is 60s
      • master + slave
      • client connects to maxscale with hintfilter and read-write-split
      • worker opens connection to maxscale on startup
      • worker is reading every 1s from maxscale
      • worker never writes (or hints) anything to the master

      Files

      • my.cnf of slave
      • maxscale.cnf
      • maxscale.log with "enable log-priority debug"

      Speculation

      I'm guessing that the connection to the master times out after 60s because it is not being used, this is somehow causing maxscale to close the connection to the client...

      the read-write split should be able to also handle that only one connection is being used

      Attachments

        1. maxscale.cnf
          1 kB
          Ludwig Gramberg
        2. maxscale-query-server-gone-away.log
          5 kB
          Ludwig Gramberg
        3. my.cnf
          1 kB
          Ludwig Gramberg

        Issue Links

          Activity

            People

              markus makela markus makela
              ludwig.gramberg Ludwig Gramberg
              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.