second boot during baremetal deploy does not configure netboot : will hang unless the machine attempts PXE automatically

[Robert Collins]

the second boot during a bare metal deploy reboots from within the deploy ramdisk rather than an IMPI 'please netboot' command, which means BIOSes that attempt local disk before netboot will fail to deploy.

[Robert Collins (lifeless)]

We're using http://serverfault.com/questions/56316/one-time-pxe-boot-from-linux-on-hp-proliant-hardware to workaround this, as booting again from PXE works. It would be nice if we could tell nova to do this (force reboot a deployed instance)

[Robert Collins (lifeless)]

The specific workaround on the hardware have is:
set system1/bootconfig1/bootsource5 bootorder=1

[aeva black (tenbrae)]

Definitely a problem!

One solution would be to remove the notify() call at the end of baremetal_deploy_helper.py: deploy(), and have the baremetal driver initiate a reboot via its power driver. That could be done in pxe.py activate_node() or driver.py just after activate_node is called.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/30378

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/30378
Committed: http://github.com/openstack/nova/commit/b00871c7e7b0d4d322e074484d481c0c48f72cd7
Submitter: Jenkins
Branch: master

commit b00871c7e7b0d4d322e074484d481c0c48f72cd7
Author: Devananda van der Veen <email address hidden>
Date: Thu May 23 18:43:59 2013 -0700

    IPMI driver sets bootdev option persistently.

    This patch changes baremetal ipmi driver's behavior with regards to
    setting the chassis boot device.

    Previous behavior: set bootdev = pxe for next boot only.
    New behavior: set bootdev = pxe persistently.

    Fixes bug 1178092

    Change-Id: I735965422a950fbd249414b5eecd0a1b21dbabea