Possible to have better error message when nbd module not loaded?

[justinsb]

When using libvirt injection, the disk image is mounted on the compute node using nbd. However, this only works is the nbd module is loaded. Otherwise, we get this error message:

WARNING nova.virt.libvirt_conn [-] instance instance-00000019: ignoring error injecting data into image 2 (nbd device /dev/nbd12 did not show up)

I think we should (1) make that message reference the fact that nbd should be loaded and/or (2) check that nbd is loaded and/or (3) automatically load it.

[Mark McLoughlin (markmc)]

Similar enough to bug #726244

Again, see:

    https://github.com/openstack/nova/blob/638ea669694f3083145e2300b362ded6749d74a1/nova/virt/disk/nbd.py#L44

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/4186

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/4186
Committed: http://github.com/openstack/nova/commit/7ffbff485e40bc3f36542f2892877adc5241accf
Submitter: Jenkins
Branch: master

commit 7ffbff485e40bc3f36542f2892877adc5241accf
Author: Pádraig Brady <email address hidden>
Date: Wed Feb 15 10:05:23 2012 +0000

    improve injection diagnostics when nbd unavailable. Bug 755854

    It's better to fail immediately with this improved diagnostic.
    Relying on qemu-nbd to detect the issue (no device), is inefficient.

    Change-Id: Ib124f6b83d8f04f0b4791c98912603f3499a9d2a