Uploaded image for project: 'Red Hat OpenStack Services on OpenShift'
  1. Red Hat OpenStack Services on OpenShift
  2. OSPRH-5294

Restart the entire Galera cluster when a cluster-incompatible config change is requested

XMLWordPrintable

    • Icon: Story Story
    • Resolution: Done
    • Icon: Undefined Undefined
    • None
    • rhos-18.0.4 (Feature Release 1)
    • mariadb-operator
    • None
    • False
    • Hide

      None

      Show
      None
    • False
    • Proposed
    • Proposed
    • Proposed
    • Proposed
    • PIDONE Board

      There are config changes in the galera CR that can't be implemented via the usual rolling update configureed by the mariadb-operator. For example:

      • Moving galera from non-TLS to TLS. Each pod must have certificated mounted before they can connect to each other.
      • mid-term: updating SST from rsync to mariabackup. Each pod must be configured to use the same SST method so they can connect to each other.
      • long-term: major upgrade of mariadb. Each node must be upgraded to the new mariadb major version before they can cluster with each other.

      Implement logics/mechanisms to stop all the galera pods before an cluster-incompatible update can be.

            rhn-engineering-dciabrin Damien Ciabrini
            rhn-engineering-dciabrin Damien Ciabrini
            rhos-dfg-pidone
            Votes:
            0 Vote for this issue
            Watchers:
            2 Start watching this issue

              Created:
              Updated:
              Resolved: