← Back to team overview

yahoo-eng-team team mailing list archive

[Bug 1939545] Re: device_path not saved into bdm.connection_info during pre_live_migration

 

Reviewed:  https://review.opendev.org/c/openstack/nova/+/804230
Committed: https://opendev.org/openstack/nova/commit/962eda94d52321c3237da870c3d0455c6f0e851b
Submitter: "Zuul (22348)"
Branch:    master

commit 962eda94d52321c3237da870c3d0455c6f0e851b
Author: Lee Yarwood <lyarwood@xxxxxxxxxx>
Date:   Wed Aug 11 12:04:11 2021 +0100

    compute: Ensure updates to bdms during pre_live_migration are saved
    
    When connecting volumes to the underlying host the virt drivers can
    attempt to stash additional metadata returned from os-brick into the
    connection_info associated with a bdm. This pretty janky behaviour
    relies on someone later calling .save() against the underlying
    BlockDeviceMapping object to persist these changes into the database as
    happens in the driver block device layer during a standard volume
    attach.
    
    However during pre_live_migration no call was made to .save() resulting
    in the changes made to the connection_info being lost. This change
    simply introduces this call at the end of the pre_live_migration method
    on the destination via the driver bdm objects we provide in
    block_device_info.
    
    Closes-Bug: #1939545
    Change-Id: Iea8896682fc28e3a5cd25afa45238272bee745e1


** 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/1939545

Title:
  device_path not saved into bdm.connection_info during
  pre_live_migration

Status in OpenStack Compute (nova):
  Fix Released

Bug description:
  Description
  ===========

  Various block based volume drivers attempt to save a device_path back
  into the stashed connection_info stored within Nova's block device
  mappings *after* connecting a volume to the underlying host. Thanks to
  the indirection caused by the various data structures used between the
  virt and compute layers this isn't actually saved into the underlying
  block device mapping database record during a typical attach flow
  until we get back into the compute and driver block device layer:

  https://github.com/openstack/nova/blob/84b61790763f91e12eebb96d955e2f83abc00d56/nova/virt/block_device.py#L613-L619

  However when an instance is being live migrated these volumes are
  connected as part of pre_live_migration on the destination and no
  attempt is made to save the updates made to the connection_info of the
  volume into the database. This isn't a massive problem as os-brick can
  for the most part lookup the device during future operations but it is
  obviously inefficient.

  This was initially hit in bug #1936439 but that bug is now being used
  to track in a trivial DEBUG log change while this bug will track in
  the underlying fix for the above issue.

  Steps to reproduce
  ==================
  * Attach an iSCSI/FC/NVME etc volume to an instance
  * Live migrate the instance
  * Confirm that device_path isn't present in the connection_info stashed in the bdm

  Expected result
  ===============
  device_path is stashed in the connection_info of the bdm

  Actual result
  =============
  device_path isn't stashed in the connection_info of the bdm

  Environment
  ===========
  1. Exact version of OpenStack you are running. See the following
    list for all releases: http://docs.openstack.org/releases/

  master

  2. Which hypervisor did you use?
     (For example: Libvirt + KVM, Libvirt + XEN, Hyper-V, PowerKVM, ...)
     What's the version of that?

  libvirt

  2. Which storage type did you use?
     (For example: Ceph, LVM, GPFS, ...)
     What's the version of that?

  LVM/iSCSI/FC/NVMe, any block based volume backends.

  3. Which networking type did you use?
     (For example: nova-network, Neutron with OpenVSwitch, ...)

  N/A

  Logs & Configs
  ==============

  See bug #1936439

To manage notifications about this bug go to:
https://bugs.launchpad.net/nova/+bug/1939545/+subscriptions



References