yahoo-eng-team team mailing list archive

[Psycho <2026831@xxxxxxxxxxxxxxxxxx>]

Public bug reported:

Description
===========

When I create an SRIOV port and attach it to an instance then
Nova/pci_devices db table for the VF is correctly updated and status of
the VF is changed from "available" to "allocated". If I detach the port
from the instance, the VF's status is also correctly reverted back to
"available".

But in case the port is deleted before it is detached from the instance,
the VF's status stays "allocated" (in the db Nova/pci_devices) and it
makes this VF unusable.


Steps to reproduce
==================


1) create an SRIOV port in Openstack (VNIC type = Direct) and attach it to a VM
2) delete the SRIOV port from Openstack without detaching it from the VM first  


Expected result
===============

1) VF detached from the VM
2) VF's status in database (Nova/pci_devices) changed to "available"


Actual result
=============

1) VF detached from the VM
2) VF's status in database (Nova/pci_devices) IS NOT changed to "available", it stays "allocated"


Environment
===========
1. Openstack version: Yoga

  rpm -qa | grep nova
  python3-novaclient-17.7.0-1.el8.noarch
  openstack-nova-conductor-25.2.0-1.el8.noarch
  python3-nova-25.2.0-1.el8.noarch
  openstack-nova-common-25.2.0-1.el8.noarch
  openstack-nova-scheduler-25.2.0-1.el8.noarch
  openstack-nova-api-25.2.0-1.el8.noarch
  openstack-nova-novncproxy-25.2.0-1.el8.noarch


2. Which hypervisor did you use?

  Libvirt + KVM

  What's the version of that?

  libvirt-7.6.0-6.el8.x86_64
  qemu-kvm-6.0.0-33.el8.x86_64
   


2. Which storage type did you use?

  This issue is storage independent.


3. Which networking type did you use?
   Neutron + openvswitch + sriovnicswitch

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

(hypervisor) nova-compute.log:

Before:

PciDevicePool(count=16,numa_node=0,product_id='XXX',tags={dev_type='type-
VF',parent_ifname='XXX',physical_network='XXX',remote_managed='false'},vendor_id='XXX')


After:

PciDevicePool(count=15,numa_node=0,product_id='XXX',tags={dev_type='type-
VF',parent_ifname='XXX',physical_network='XXX',remote_managed='false'},vendor_id='XXX')

** Affects: nova
     Importance: Undecided
         Status: New

-- 
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/2026831

Title:
  Table nova/pci_devices is not updated after removing attached SRIOV
  port

Status in OpenStack Compute (nova):
  New

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

  When I create an SRIOV port and attach it to an instance then
  Nova/pci_devices db table for the VF is correctly updated and status
  of the VF is changed from "available" to "allocated". If I detach the
  port from the instance, the VF's status is also correctly reverted
  back to "available".

  But in case the port is deleted before it is detached from the
  instance, the VF's status stays "allocated" (in the db
  Nova/pci_devices) and it makes this VF unusable.


  Steps to reproduce
  ==================

  
  1) create an SRIOV port in Openstack (VNIC type = Direct) and attach it to a VM
  2) delete the SRIOV port from Openstack without detaching it from the VM first  


  Expected result
  ===============

  1) VF detached from the VM
  2) VF's status in database (Nova/pci_devices) changed to "available"

  
  Actual result
  =============

  1) VF detached from the VM
  2) VF's status in database (Nova/pci_devices) IS NOT changed to "available", it stays "allocated"


  Environment
  ===========
  1. Openstack version: Yoga

    rpm -qa | grep nova
    python3-novaclient-17.7.0-1.el8.noarch
    openstack-nova-conductor-25.2.0-1.el8.noarch
    python3-nova-25.2.0-1.el8.noarch
    openstack-nova-common-25.2.0-1.el8.noarch
    openstack-nova-scheduler-25.2.0-1.el8.noarch
    openstack-nova-api-25.2.0-1.el8.noarch
    openstack-nova-novncproxy-25.2.0-1.el8.noarch

  
  2. Which hypervisor did you use?

    Libvirt + KVM

    What's the version of that?

    libvirt-7.6.0-6.el8.x86_64
    qemu-kvm-6.0.0-33.el8.x86_64
     


  2. Which storage type did you use?

    This issue is storage independent.


  3. Which networking type did you use?
     Neutron + openvswitch + sriovnicswitch

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

  (hypervisor) nova-compute.log:

  Before:

  PciDevicePool(count=16,numa_node=0,product_id='XXX',tags={dev_type='type-
  VF',parent_ifname='XXX',physical_network='XXX',remote_managed='false'},vendor_id='XXX')

  
  After:

  PciDevicePool(count=15,numa_node=0,product_id='XXX',tags={dev_type='type-
  VF',parent_ifname='XXX',physical_network='XXX',remote_managed='false'},vendor_id='XXX')

To manage notifications about this bug go to:
https://bugs.launchpad.net/nova/+bug/2026831/+subscriptions