We can't *update* an undercloud when the overcloud is in failed state.

[Sofer Athlan-Guyot]

Hi,

originally reported there https://bugzilla.redhat.com/show_bug.cgi?id=1655629

The problem of $title is that if we need a patch in the mistral workbook for instance to make the overcloud switch from error to succes, then we can't as the update of the mistral workbooks happen at the end of an undercloud update.

[Sofer Athlan-Guyot (sofer-athlan-guyot)]

Note, still unsure about how to do a Queen only lp.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: stable/queens
Review: https://review.openstack.org/622290

[Juan Antonio Osorio Robles (juan-osorio-robles)]

Is this still an issue?

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: stable/queens
Review: https://review.openstack.org/649970

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/622290
Committed: https://git.openstack.org/cgit/openstack/instack-undercloud/commit/?id=a96812f88ac6ff604c78ca78b312deb847ecec27
Submitter: Zuul
Branch: stable/queens

commit a96812f88ac6ff604c78ca78b312deb847ecec27
Author: Sofer Athlan-Guyot <email address hidden>
Date: Tue Dec 4 13:15:17 2018 +0100

    Enable override of overcloud status check for undercloud update.

    We need to migrate heat to a new engine during pike to queen upgrade
    and that requires an overcloud in a sane state.

    The problem is that during undercloud update (from queen to latest
    queen) we may need to update the undercloud even if the overcloud is
    in a failed state. For instance if we need a fix in a mistral
    workbook to make an overcloud action successful.

    This patch offers the possibility to by-pass the overcloud check in
    such case.

    A simple environment variable setting is used to communicate with the
    python script triggered by instack-pre-upgrade-undercloud et
    instack-upgrade-undercloud. It looks cleaner that mangling with the
    python -c line in the scripts.

    So if the environment variable TRIPLEO_FORCED_UPDATE is set to
    non-empty string or if TRIPLEO_FORCED_UPDATE is passed as an argument
    of the two scripts mentioned above then we trigger the no-check heat
    stack status mode.

    Change-Id: I43a1b4892f25bb649ee34d39b5fa462a302834aa
    Partial-Bug: #1806683

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/649970
Committed: https://git.openstack.org/cgit/openstack/python-tripleoclient/commit/?id=6ce2039d37085a03da036c8ec6cc6ef223bfb04c
Submitter: Zuul
Branch: stable/queens

commit 6ce2039d37085a03da036c8ec6cc6ef223bfb04c
Author: Sofer Athlan-Guyot <email address hidden>
Date: Thu Apr 4 13:12:55 2019 +0200

    [Queen Only] Add force option to undercloud update.

    When the overcloud is in a failed state we can't run the undercloud
    upgrade command anymore. But we may need to inject code in mistral for
    instance to get out of the overcloud failure.

    The force option enables the user to get out of this loop.

    Queen only as instack undercloud is removed from Rocky on.

    Change-Id: I9e7fe8cfd96262938d463de0d6c45ccedc9989be
    Closes-Bug: #1806683
    Depends-On: https://review.openstack.org/622290

[wes hayutin (weshayutin)]

both patches have merged, moving to fixed_released, reopen if needed.

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/python-tripleoclient 9.3.0 release.