yahoo-eng-team team mailing list archive

[Dvir Saidof <1836065@xxxxxxxxxxxxxxxxxx>]

Public bug reported:

Initially, the problem was that I couldn't attach more than 25 volumes
to a single VM. I would receive the following error in the "nova-
compute.log" when trying to attach the 26th volume.

ERROR nova.virt.libvirt.driver [req-39ac33eb-e924-4f51-ae7a-e2847505f756
d169224fc7764363895d00fe9d13fb16 2fb563a2af214ebb90efd30f0033de66 -
default default] [instance: 593b3e25-4c02-40dd-acc9-6ee554b0bdd2] Failed
to attach volume at mountpoint: /dev/vd{: libvirtError: XML error:
Unknown disk name 'vd{' and no address specified

This issue has been discussed and solved not long ago and I overcame that and managed to attach the 26th volume with id "vdaa" backporting from here: 
(1) https://review.opendev.org/#/c/631166/
(2) https://review.opendev.org/#/c/616777/

The problem now is that I've got a new error, trying to attach the 27th volume. This is the error I get in "nova-compute.log":
2019-07-03 09:30:55.657 8 ERROR oslo_messaging.rpc.server libvirtError: internal error: No more available PCI slots


Steps to reproduce on rocky version:
- update block_device.py, utils.py, compute.py and blockinfo.py on nova-compute docker from (1) and (2).
- attach 27 volumes to a VM

Expected results: 
- 27 volumes being attached

Actual result:
- only 26 attached to the VM and the following error on nova-compute.log
2019-07-03 09:30:55.657 8 ERROR oslo_messaging.rpc.server libvirtError: internal error: No more available PCI slots

** 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/1836065

Title:
  libvirt Error - No more available PCI slots

Status in OpenStack Compute (nova):
  New

Bug description:
  Initially, the problem was that I couldn't attach more than 25 volumes
  to a single VM. I would receive the following error in the "nova-
  compute.log" when trying to attach the 26th volume.

  ERROR nova.virt.libvirt.driver [req-39ac33eb-e924-4f51-ae7a-
  e2847505f756 d169224fc7764363895d00fe9d13fb16
  2fb563a2af214ebb90efd30f0033de66 - default default] [instance:
  593b3e25-4c02-40dd-acc9-6ee554b0bdd2] Failed to attach volume at
  mountpoint: /dev/vd{: libvirtError: XML error: Unknown disk name 'vd{'
  and no address specified

  This issue has been discussed and solved not long ago and I overcame that and managed to attach the 26th volume with id "vdaa" backporting from here: 
  (1) https://review.opendev.org/#/c/631166/
  (2) https://review.opendev.org/#/c/616777/

  The problem now is that I've got a new error, trying to attach the 27th volume. This is the error I get in "nova-compute.log":
  2019-07-03 09:30:55.657 8 ERROR oslo_messaging.rpc.server libvirtError: internal error: No more available PCI slots

  
  Steps to reproduce on rocky version:
  - update block_device.py, utils.py, compute.py and blockinfo.py on nova-compute docker from (1) and (2).
  - attach 27 volumes to a VM

  Expected results: 
  - 27 volumes being attached

  Actual result:
  - only 26 attached to the VM and the following error on nova-compute.log
  2019-07-03 09:30:55.657 8 ERROR oslo_messaging.rpc.server libvirtError: internal error: No more available PCI slots

To manage notifications about this bug go to:
https://bugs.launchpad.net/nova/+bug/1836065/+subscriptions