VM get stuck when pxe_ssh is deployed with local boot

[Lucas Alvares Gomes]

This problem only affects the DIB ramdisk.

When deploying a machine with local boot using the pxe_ssh driver the VM will keep rebooting after the deployment and won't boot into the image written onto local disk. This happens because Ironic will change the boot device from PXE to DISK as part of the deployment but the change will only take effect after the VM is powered off and on again (that's a virsh problem).

When deploying with the DIB ramdisk, at the end of the deployment the ramdisk itself will reboot the machine (not power off and on), so on the next boot the VM will try to boot from the network again and will fail.

My suggestion is to have Ironic to control the power state of the machine after the deployment. Once the image is written to the disk and the bootloader installed (in case local boot is enabled) Ironic should do the reboot.

Workaround:

To workaround this problem one can power the VM off and on again manually and then it will boot from the local disk, e.g:

$ virsh destroy <machine name>
$ virsh start <machine name>

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/170117

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/170117
Committed: https://git.openstack.org/cgit/openstack/ironic/commit/?id=af625b8b19567ff8618ab9f0a0ba76c086a5042f
Submitter: Jenkins
Branch: master

commit af625b8b19567ff8618ab9f0a0ba76c086a5042f
Author: Lucas Alvares Gomes <email address hidden>
Date: Thu Apr 2 14:14:39 2015 +0100

    Fix VM stuck when deploying with pxe_ssh + local boot

    When deploying a machine with the DIB ramdisk, pxe_ssh driver and
    local boot the VM will keep rebooting forever after the deployment is
    completed. This happens because Ironic will change the boot device from
    PXE to DISK as part of the deployment but the change will only take effect
    after the VM is powered off and on again (that's a virsh problem). This
    patch is fixing this problem by making Ironic reboot the machine after
    the deployment is completed (reboot in Ironic means power it off and
    on again).

    Closes-Bug: #1439664
    Change-Id: I4aa859b00a8bba186e920e09734ec2e4c45dec3e