state machine incorrectly models CLEANFAIL -> CLEANING transition

[aeva black]

The state machine incorrectly models the transitions around the CLEANFAIL state.

For reference, I'll quote the spec:
  http://specs.openstack.org/openstack/ironic-specs/specs/kilo/new-ironic-state-machine.html#proposed-change

CLEANING
    Nodes in the CLEANING state are being scrubbed in preparation to being made AVAILABLE. Good candidates for CLEANING tasks include:
    [snip...]
    Management of CLEANING tasks should be handled in the same fashion as ZAPPING tasks.

ZAPFAIL
    Nodes that transition into ZAPFAIL will automatically enter maintenance mode, as failure to ZAP a machine usually indicates a hardware failure or something else that requires remote hands to fix.

Here is the incorrect code:

204 # Add clean* states
205 machine.add_state(CLEANING, target=AVAILABLE, **watchers)
206 machine.add_state(CLEANFAIL, target=AVAILABLE, **watchers)
...
262 # If cleaning fails, wait for operator intervention
263 machine.add_transition(CLEANING, CLEANFAIL, 'fail')
264
265 # A node that fails cleaning may be put back through cleaning
266 machine.add_transition(CLEANFAIL, CLEANING, 'clean')
267
268 # An operator may want to hold a CLEANFAIL node in operator for zapping or
269 # outside-of-Ironic operations (like replacing hardware)
270 machine.add_transition(CLEANFAIL, MANAGEABLE, 'manage')

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/176032

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/176032
Committed: https://git.openstack.org/cgit/openstack/ironic/commit/?id=0c295153af55f85120c78875052bf5c23e168c4b
Submitter: Jenkins
Branch: master

commit 0c295153af55f85120c78875052bf5c23e168c4b
Author: Ruby Loo <email address hidden>
Date: Tue Apr 21 19:07:32 2015 +0000

    Remove state transition: CLEANFAIL -> CLEANING

    The state machine incorrectly models the transition from CLEANFAIL to
    to CLEANING state. The spec [1] mentions that management of CLEANING
    tasks should be handled in the same fashion as ZAPPING tasks. For
    ZAPFAIL:
        Nodes that transition into ZAPFAIL will automatically enter
        maintenance mode, as failure to ZAP a machine usually indicates
        a hardware failure or something else that requires remote hands
        to fix.

    Nodes in CLEANFAIL should do a similar thing as ZAPFAIL; they are
    put in maintenance mode, but shouldn't be able to transition directly
    to CLEANING.

    This patch also clarifies the documentation for the transition
    CLEANFAIL -> MANAGEABLE.

    [1] http://specs.openstack.org/openstack/ironic-specs/specs/kilo/new-ironic-state-machine.html#proposed-change

    Change-Id: Ieda2fcd78332a063c51685cd01bb824014f0f4a8
    Closes-Bug: #1446758