yahoo-eng-team team mailing list archive

[OpenStack Infra <1835822@xxxxxxxxxxxxxxxxxx>]

Reviewed:  https://review.opendev.org/669738
Committed: https://git.openstack.org/cgit/openstack/nova/commit/?id=641feb62a33e211a3ed1f9a2309ee16d827f5fe2
Submitter: Zuul
Branch:    master

commit 641feb62a33e211a3ed1f9a2309ee16d827f5fe2
Author: Sean Mooney <work@xxxxxxxxxxxxxxx>
Date:   Mon Jul 8 18:35:11 2019 +0000

    make config drives sticky bug 1835822
    
    This change reorders the call in
    _update_instance_after_spawn so that we call
    configdrive.update_instance before we set
    instance.launched_at. This ensures that if the vm
    is booted with a config drive because the host had
    force_config_drive=true the instance will keep its
    config drive across reboots.
    
    This change fixes a regression introduced as part
    of fixing bug #1827492 which addressed failing
    to boot vms after changing force_config_drive=false
    to force_config_drive=true.
    
    Change-Id: I9194423f5f95e9799bd891548e24756131d65e76
    Related-Bug: #1827492
    Closes-Bug: #1835822


** Changed in: nova
       Status: In Progress => Fix Released

-- 
You received this bug notification because you are a member of Yahoo!
Engineering Team, which is subscribed to OpenStack Compute (nova).
https://bugs.launchpad.net/bugs/1835822

Title:
  vms loose acess to config drive with CONF.force_config_drive=True
  after hard reboot

Status in OpenStack Compute (nova):
  Fix Released

Bug description:
  The fix to bug https://bugs.launchpad.net/nova/+bug/1827492
  https://review.opendev.org/#/c/659703/8

  changed the behavior of nova.virt.configdrive.required_by
  to depend on instance.launched_at

  https://review.opendev.org/#/c/659703/8/nova/virt/configdrive.py@196

  but did not reorder
  https://github.com/openstack/nova/blob/86524773b8cd3a52c98409c7ca183b4e1873e2b8/nova/compute/manager.py#L1757-L1758

  as a result when nova.compute.manager._update_instance_after_spawn
  is called instance.launched_at is always set before we call nova.virt.configdrive.update_instance

  as a result instance.config_drive will always be set to false if not
  set on the api.

  this results in a vm that is spawned on a host with force_config_drive=True initally spawning
  with a config drive but loosing it after a hard reboot.

  for any deployment that uses config driver for vendor data or device
  role tagging because they do not deploy the metadata service this is a
  regressions as they cannot fall back to the metadta service.

  this also might cause issue for deployment that support the deprectated file injection api that is part of the v2.1 api as the files are only stored in the config drive and are not part metadta endoint
  note: i have not checked if we autoset instnace.config_drive when you use file injection or not so it may be unaffected since the breakage of the other support uscases is enough to justify this bug.

  
  the fix is simple jsut swap the order of https://github.com/openstack/nova/blob/86524773b8cd3a52c98409c7ca183b4e1873e2b8/nova/compute/manager.py#L1757-L1758

  and then instance will have there instnace.config_drive value set correctly when they first boot
  and it will be sticky for the lifetime of the instance.

To manage notifications about this bug go to:
https://bugs.launchpad.net/nova/+bug/1835822/+subscriptions