Uploaded image for project: 'MariaDB MaxScale'
  1. MariaDB MaxScale
  2. MXS-2342

maxadmin commands hang when master pod deleted after failover occurs

    XMLWordPrintable

Details

    • Bug
    • Status: Closed (View Workflow)
    • Major
    • Resolution: Fixed
    • 2.3.3
    • 2.3.5
    • mariadbmon, maxadmin
    • None
    • kubernetes openstack deployment

    Description

      All maxadmin commands seem to hang for 30 to over 60 seconds during pod failover using mariadbmon. The following are just two of the commands that seem to hang:
      maxadmin list servers (which we use to determine server status)
      maxadmin show monitor MariaDB-Monitor (used to audit auto-failover flag).

      I have no idea what maxscale itself is doing during this time. Our monitors running on this container will eventually detect that the IP has changed for the deleted pod and perform a maxadmin alter server to update the IP address, but that doesn't happen for quite a while after the pod is recreated due to all the hung maxadmin commands blocking this from happening.

      This does NOT happen on maxscale-2.2.19.

      Attached is:
      maxscale.log - maxscale logs during this time
      ps.out - ps -ef on the container with date every 5 minutes to show hung maxadmin processes
      maxadmin.time - shows the running time of a couple manual maxadmin list servers

      Attachments

        1. maxadmin.time
          1 kB
        2. maxscale.log
          59 kB
        3. ps.out
          64 kB
        4. maxscale.cnf
          2 kB

        Activity

          People

            esa.korhonen Esa Korhonen
            rvlane Richard Lane
            Votes:
            0 Vote for this issue
            Watchers:
            2 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.