Comment 24 for bug 2025480

Revision history for this message
OpenStack Infra (hudson-openstack) wrote : Fix merged to nova (stable/2023.1)

Reviewed: https://review.opendev.org/c/openstack/nova/+/904373
Committed: https://opendev.org/openstack/nova/commit/683ecc060e3bca818b9fb514d297e323bc8cb220
Submitter: "Zuul (22348)"
Branch: stable/2023.1

commit 683ecc060e3bca818b9fb514d297e323bc8cb220
Author: Bence Romsics <email address hidden>
Date: Wed Aug 2 16:22:55 2023 +0200

    Do not untrack resources of a server being unshelved

    This patch concerns the time when a VM is being unshelved and the
    compute manager set the task_state to spawning, claimed resources of
    the VM and then called driver.spawn(). So the instance is in vm_state
    SHELVED_OFFLOADED, task_state spawning.

    If at this point a new update_available_resource periodic job is
    started that collects all the instances assigned to the node to
    calculate resource usage. However the calculation assumed that a
    VM in SHELVED_OFFLOADED state does not need resource allocation on
    the node (probably being removed from the node as it is offloaded)
    and deleted the resource claim.

    Given all this we ended up with the VM spawned successfully but having
    lost the resource claim on the node.

    This patch changes what we do in vm_state SHELVED_OFFLOADED, task_state
    spawning. We no longer delete the resource claim in this state and
    keep tracking the resource in stats.

    Change-Id: I8c9944810c09d501a6d3f60f095d9817b756872d
    Closes-Bug: #2025480
    (cherry picked from commit f1dc4ec39bcfda1bd4b97e233a9da498b6378c4f)
    (cherry picked from commit 4239d1fec2814c074482b740a2fd38a5d5ce6942)