[openstack-dev] [Ironic] A strange transition in Ironic FSM

Vladyslav Drok vdrok at mirantis.com
Mon Feb 8 12:28:00 UTC 2016


Hi all,

Looking at the state machine spec
<https://specs.openstack.org/openstack/ironic-specs/specs/kilo-implemented/new-ironic-state-machine.html>,
this might be a leftover from the older state machine.
When the node is in deleting state, we only clean up the kernel/ramdisk and
clear DHCP
options, so rebuild may be completed successfully afterwards. I agree that
this transition
does not make much sense, and removing it will help fixing the bug you've
referenced,
but I'm not sure how to handle backwards compatibility in this case (and
whether we need
to do this).

Vlad

On Fri, Feb 5, 2016 at 4:00 PM, Yuriy Zveryanskyy <yzveryanskyy at mirantis.com
> wrote:

> Hi.
>
> We have a followed transition in common/states.py:
>
> # An errored instance can be rebuilt
> # ironic/conductor/manager.py:do_node_deploy()
> machine.add_transition(ERROR, DEPLOYING, 'rebuild')
>
> At first glance it looks correct. But ERROR state is
> used only for error after deleting, see
> http://docs.openstack.org/developer/ironic/_images/states.svg
> So ERROR is delete error, at least now, and transition
> error (delete error) -> deploying (on_rebuild)
> is possible.
> Looks strange if operator wants to remove an instance completely and then
> does rebuild after error (non-error targets for deleting is cleaning ->
> available).
> I think this transition should be removed. Without this strange transition
> bug https://bugs.launchpad.net/ironic/+bug/1522008 can be fixed by simple
> way,
> port's vif id can be removed via Ironic virt driver request before waiting
> of CLEANING
> (it's no more needed).
>
> Yuriy Zveryanskyy
>
>
>
>
>
>
> __________________________________________________________________________
> OpenStack Development Mailing List (not for usage questions)
> Unsubscribe: OpenStack-dev-request at lists.openstack.org?subject:unsubscribe
> http://lists.openstack.org/cgi-bin/mailman/listinfo/openstack-dev
>
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://lists.openstack.org/pipermail/openstack-dev/attachments/20160208/c0358a2b/attachment.html>


More information about the OpenStack-dev mailing list