Baremetal pxe driver doesn't delete /tftpboot/{uuid}/*

[aeva black]

Baremetal PXE driver deactivate_bootloader() is failing to delete tftp images when driver.destroy() is called. This results in a non-empty /tftpboot/{instance-uuid}/ directory being left behind on the compute node after an instance is deleted.

I believe this happens because the instance parameter passed to driver.destroy() does not contain ['extra_specs']['deploy_kernel_id'] or ['extra_specs']['deploy_ramdisk_id']. These these were passed to driver.spawn(), so I had assumed they were passed to driver.destroy() as well. This can be confirmed by logging the exception within the first try: block of deactivate_bootloader().

One solution is to pass extra_specs to driver.destroy().
Another solution is for deactivate_bootloader() to blindly delete all files under the /tftpboot/{instance-uuid}/ directory.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/20039

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/20039
Committed: http://github.com/openstack/nova/commit/e306beac9297f4311aa3989e93b0bd19eb4d7c77
Submitter: Jenkins
Branch: master

commit e306beac9297f4311aa3989e93b0bd19eb4d7c77
Author: Devananda van der Veen <email address hidden>
Date: Fri Jan 18 09:40:37 2013 -0800

    PXE driver should rmtree directories it created

    Baremetal PXE driver was failing to delete the per-instance tftpboot and
    image directories which it created when the instance was deleted. This
    happened partly because of dangling files within the directory, and
    partly because 'unlink' does not remove directories and the error was
    squelched.

    Now, when destroy()ing an instance, PXE driver will call shutil.rmtree()
    on the per-instance directories which it created.

    Fixes bug 1101048.

    Change-Id: I682d0264288add488ea23e5d5200357b7be52dd9