Uploaded image for project: 'Red Hat OpenStack Services on OpenShift'
  1. Red Hat OpenStack Services on OpenShift
  2. OSPRH-17141

RHOSP17 compute node booted with instance disk

XMLWordPrintable

    • 3
    • False
    • Hide

      None

      Show
      None
    • False
    • ?
    • openstack-tripleo-image-elements-13.1.3-17.1.20250728121644.a641940.el9ost
    • openstack-tripleo-image-elements-13.1.3-17.1.20250728121644.a641940.el9ost
    • Impediment
    • rhos-ops-day1day2-hardprov
    • None
    • Hide
      Cause:

      Duplicate disk labels were used for Octavia images and Overcloud OS node images. In some cases where external SAN storage was used, an Octavia image would be mounted instead of the root OS for the Overcloud node.

      Consequence:

      The Overcloud Compute node would boot to the wrong operating system image and would be unable to operate as a compute node until booted to the correct OS image.

      Fix:

      The default disk image label for compute nodes has been updated to a new default label. This prevents this issue from occuring on new deployments.

      Result:

      New deployments utilizing the updated image should not fall into this fall into the failure case with external SAN storage in use.
      Show
      Cause: Duplicate disk labels were used for Octavia images and Overcloud OS node images. In some cases where external SAN storage was used, an Octavia image would be mounted instead of the root OS for the Overcloud node. Consequence: The Overcloud Compute node would boot to the wrong operating system image and would be unable to operate as a compute node until booted to the correct OS image. Fix: The default disk image label for compute nodes has been updated to a new default label. This prevents this issue from occuring on new deployments. Result: New deployments utilizing the updated image should not fall into this fall into the failure case with external SAN storage in use.
    • Bug Fix
    • HardProv Sprint 7, HardProv Sprint 9, HardProv Sprint 14
    • 3
    • Moderate

      To Reproduce Steps to reproduce the behavior:

      1. Deploy a RHOSP17.1.4 with PURE storage over fiber channel for volumes and compute.
      2. Create a instance.
      3. Reboot the compute node where are the instance.
      4. See that the compute node booted from the instance volume.

      Expected behavior

      • Compute node boots from it's disk only.

      Device Info (please complete the following information):

      • Hardware Specs: Instance volumes on PURE storage over fiber channel.
      • OS Version: RHEL 9.2

      Bug impact

      • Lost a compute at it reboot.

      Known workaround

      • Replace the kernel argument boot=LABEL=$label with boot=UUID=$partition_uuid

      Additional context

      • Compute node kernel parameter set as "boot=LABEL=$label" (the default here) and instance have the same $label.
      • the root disk in ironic property is rotational=true.

              jkreger@redhat.com Julia Kreger
              rh-ee-cgussobo Conrado Gusso Bozza
              rhos-dfg-hardprov
              Votes:
              0 Vote for this issue
              Watchers:
              7 Start watching this issue

                Created:
                Updated: