Uploaded image for project: 'OpenShift Bugs'
  1. OpenShift Bugs
  2. OCPBUGS-9996

CSI driver ends up in crash loop when host does not have default VG name

    XMLWordPrintable

Details

    • Bug
    • Resolution: Done-Errata
    • Minor
    • 4.13.0, 4.14.0
    • 4.14.0
    • MicroShift
    • None
    • No
    • uShift Sprint 233
    • 1
    • Rejected
    • False
    • Hide

      None

      Show
      None
    • Hide
      Made the CSI driver default configuration more robust. Before this change, unless the CSI driver was explicitly configured the host needed to have a volume group named "rhel" in order for the driver to work. Now, microshift examines the volume groups that exist to pick one as the default. If there is only 1 volume group, it is used. If there are multiple volume groups and one is named "microshift", it is used. Otherwise, the CSI driver must be explicitly configured to be enabled.
      Show
      Made the CSI driver default configuration more robust. Before this change, unless the CSI driver was explicitly configured the host needed to have a volume group named "rhel" in order for the driver to work. Now, microshift examines the volume groups that exist to pick one as the default. If there is only 1 volume group, it is used. If there are multiple volume groups and one is named "microshift", it is used. Otherwise, the CSI driver must be explicitly configured to be enabled.
    • Enhancement
    • Done

    Description

      Description of problem:

      Running MicroShift on a system without an lvmd.conf file can end up with the CSI containers in a crash-loop state if the host is not configured with a volume group named "rhel". Since that is not the default name for the volume group when installing RHEL via Anaconda, many users may experience this situation until they explicitly configure the driver.

      Version-Release number of selected component (if applicable):

      4.12 and later

      How reproducible:

      Fairly often

      Steps to Reproduce:

      1. Deploy a stock RHEL RPM-based system without taking any special steps to name the volume group.
      2. dnf install microshift
      3. systemctl start microshift
      

      Actual results:

      The pod running lvmd will crash loop.

      Expected results:

      No crashing pods.

      Additional info:

       

      Attachments

        Issue Links

          Activity

            People

              dhellman@redhat.com Doug Hellmann
              dhellman@redhat.com Doug Hellmann
              Patryk Matuszak Patryk Matuszak
              Votes:
              0 Vote for this issue
              Watchers:
              5 Start watching this issue

              Dates

                Created:
                Updated:
                Resolved: