id: PG-22
category: vm-lifecycle
title: Resize not finished?
situation: >-
  After a resize or a cold migration, the instance sits in `VERIFY_RESIZE`, `RESIZE` or `ERROR`. Three
  states, three different situations.

diagram:
  type: branches
  head: [Status and task state, It means]
  split: 46
  rows:
    - see: "`VERIFY_RESIZE`, no task state"
      means: It worked
      sub: It waits for a confirm or a revert
      tone: ok
    - see: "`RESIZE`, with a task state"
      means: It stopped part way
      sub: A service died, or a message was lost
      tone: warn
    - see: "`ERROR`, no task state"
      means: A step failed
      sub: The fault says which one
      tone: stop
  note: >-
    In `VERIFY_RESIZE` the old copy is kept: Placement holds resources on both computes until the confirm.

checks:
  - label: State, task state, fault
    hint: openstack server show <id> -c status -c OS-EXT-STS:task_state -c fault
  - label: The migration record
    hint: openstack server migration list --server <id>
  - label: Its actions, then the failed step
    hint: "openstack server event list <id>, then event show <id> <req-id>"
  - label: Are both computes up?
    hint: openstack compute service list --service nova-compute

caution:
  - level: destructive
    text: >-
      Confirm deletes the old copy on the source. Revert loses what the guest wrote to its local disks since
      the resize. Both are the owner's call.
  - level: avoid
    text: >-
      `server set --state active` on an instance in `ERROR`. It changes the database only: find out where the
      disks are first.

lesson: VERIFY_RESIZE is not stuck. It waits for a decision only the owner can make.

runbook:
  path: lifecycle/resize-stuck-in-verify-resize-or-failed.md
  status: lab-verified
  release: "2026.1"

scope: >-
  Written from the Nova 2026.1 source and lab-verified with Kolla-Ansible, two computes and local qcow2
  disks: the unconfirmed resize, a service dying part way, SSH between computes and a smaller disk were
  reproduced; a destination that cannot run the guest was not. Nova does not shrink disks, except a root
  disk booted from volume. With resize_confirm_window set, resizes older than the window are confirmed without
  the owner and can no longer be reverted.

linkedin:
  title: "VERIFY_RESIZE is not a stuck state. It is a pending decision."
  caption: |
    After a resize or a cold migration, an instance can sit in three states that mean very different things.

    → VERIFY_RESIZE, no task state: the resize worked. It waits for a confirm or a revert, and can wait for days.
    → RESIZE, with a task state: the operation stopped part way. A service died, or a message was lost.
    → ERROR: a step failed, and the fault says which one.

    The first is the one that surprises people. While an instance stays in VERIFY_RESIZE, the old copy is kept on the source and Placement holds resources on both computes. Many of them make hosts look fuller than they are.

    And neither way out can be undone: confirm deletes the old copy, revert loses what the guest wrote to its local disks since the resize. That decision belongs to the owner.
  hashtags: [OpenStack, Nova, Placement, CloudOperations, SRE, OzarLand]
