Confusing error message when trying to disable non-nova-compute service

[Jack Ivanov]

I'm trying to disable nova services, but the requests fail.

python-nova-17.0.7-1.el7.noarch
openstack-nova-api-17.0.7-1.el7.noarch
openstack-nova-novncproxy-17.0.7-1.el7.noarch
openstack-nova-placement-api-17.0.7-1.el7.noarch
python2-novaclient-10.1.0-1.el7.noarch
openstack-nova-conductor-17.0.7-1.el7.noarch
openstack-nova-scheduler-17.0.7-1.el7.noarch
openstack-nova-console-17.0.7-1.el7.noarch
openstack-nova-common-17.0.7-1.el7.noarch

How to reproduce:
# openstack compute service set p024.domain.com nova-consoleauth --disable
Failed to set service status to disabled
Compute service nova-consoleauth of host p024.domain.com failed to set.

Logs:
Action: 'update', calling method: <function version_select at 0x7f49474a8938>, body: {"binary": "nova-consoleauth", "host": "p024.domain.com"} _process_stack /usr/lib/python2.7/site-packages/nova/api/openstack/wsgi.py:604
HTTP exception thrown: Host 'p024.domain.com' is not mapped to any cell
Returning 404 to user: Host 'p024.domain.com' is not mapped to any cell __call__ /usr/lib/python2.7/site-packages/nova/api/openstack/wsgi.py:1064
XX.XX.XX.XX "PUT /v2.1/os-services/disable HTTP/1.1" status: 404 len: 511 time: 0.3039389

Why nova-consoleauth should be mapped to a cell? Is that a bug?

Thanks!

[Jack Ivanov (gunph1ld)]

# nova-manage cell_v2 simple_cell_setup
Cell0 is already setup
All hosts are already mapped to cell(s), exiting.

[Takashi Natsume (natsume-takashi)]

The "PUT /os-services/disable" API is just for a compute service (nova-compute).
It says that "Disables scheduling for a Compute service." in the compute API reference.

https://developer.openstack.org/api-ref/compute/?expanded=disable-scheduling-for-a-compute-service-detail#disable-scheduling-for-a-compute-service

But the error message is confusing and there are no checks whether the 'binary' is nova-compute or not.
It had better be fixed.

[Jack Ivanov (gunph1ld)]

So there is no way to disable other services? I would love to see that implemented

[Matt Riedemann (mriedem)]

As Takashi noted, disabling/enabling/forcing down/up services is only for compute services. There is no functional point to disabling conductor/scheduler/consoleauth. This was noted in the Pike release notes:

https://docs.openstack.org/releasenotes/nova/pike.html#relnotes-16-0-0-stable-pike-other-notes

The thing I think Takashi notes is the confusing error message:

"HTTP exception thrown: Host 'p024.domain.com' is not mapped to any cell"

We could check the binary in that case and if it's not 'compute' then we can provide a more useful error message.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/620667

[Jack Ivanov (gunph1ld)]

This is not only about the error message. The command line client also should be fixed.

# openstack help compute service set
usage: openstack compute service set [-h] [--enable | --disable]
                                     [--disable-reason <reason>]
                                     [--up | --down]
                                     <host> <service>
...

It's confusing and absolutely doesn't make any sense of specifying <service> in the query, because there is no any choice to specify anything except nova-compute.

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/620667
Committed: https://git.openstack.org/cgit/openstack/nova/commit/?id=95cbc35b0bdae0de15569bc35fadbcf970da5b9e
Submitter: Zuul
Branch: master

commit 95cbc35b0bdae0de15569bc35fadbcf970da5b9e
Author: Matt Riedemann <email address hidden>
Date: Wed Nov 28 12:52:17 2018 -0500

    Provide a useful error message when trying to update non-compute services

    Starting in Pike, we disallowed trying to update (enable/disable/force down)
    non-nova-compute services because of multi-cell support using host mappings
    to lookup service records, and simply because disabling non-compute services
    doesn't do anything.

    However, before microversion 2.53, the error the user gets back is confusing:

      HTTP exception thrown: Host 'p024.domain.com' is not mapped to any cell

    This change provides a useful error message in this case and also changes
    the 404 response to a 400 response to align with the type of error and the
    behavior of the 2.53 microversion.

    Change-Id: I44f09aec60b0b18c458f9ba6d8b725db962e9cc7
    Closes-Bug: #1805164

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: stable/rocky
Review: https://review.openstack.org/636955

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/636955
Committed: https://git.openstack.org/cgit/openstack/nova/commit/?id=315d9fab7aaa6b9835e80f5f56b49b44d382ca1f
Submitter: Zuul
Branch: stable/rocky

commit 315d9fab7aaa6b9835e80f5f56b49b44d382ca1f
Author: Matt Riedemann <email address hidden>
Date: Wed Nov 28 12:52:17 2018 -0500

    Provide a useful error message when trying to update non-compute services

    Starting in Pike, we disallowed trying to update (enable/disable/force down)
    non-nova-compute services because of multi-cell support using host mappings
    to lookup service records, and simply because disabling non-compute services
    doesn't do anything.

    However, before microversion 2.53, the error the user gets back is confusing:

      HTTP exception thrown: Host 'p024.domain.com' is not mapped to any cell

    This change provides a useful error message in this case and also changes
    the 404 response to a 400 response to align with the type of error and the
    behavior of the 2.53 microversion.

    Change-Id: I44f09aec60b0b18c458f9ba6d8b725db962e9cc7
    Closes-Bug: #1805164
    (cherry picked from commit 95cbc35b0bdae0de15569bc35fadbcf970da5b9e)

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: stable/queens
Review: https://review.openstack.org/638803

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/638803
Committed: https://git.openstack.org/cgit/openstack/nova/commit/?id=b041160c8d56f657059743bc5a0e6dbefc27ddd6
Submitter: Zuul
Branch: stable/queens

commit b041160c8d56f657059743bc5a0e6dbefc27ddd6
Author: Matt Riedemann <email address hidden>
Date: Wed Nov 28 12:52:17 2018 -0500

    Provide a useful error message when trying to update non-compute services

    Starting in Pike, we disallowed trying to update (enable/disable/force down)
    non-nova-compute services because of multi-cell support using host mappings
    to lookup service records, and simply because disabling non-compute services
    doesn't do anything.

    However, before microversion 2.53, the error the user gets back is confusing:

      HTTP exception thrown: Host 'p024.domain.com' is not mapped to any cell

    This change provides a useful error message in this case and also changes
    the 404 response to a 400 response to align with the type of error and the
    behavior of the 2.53 microversion.

    Change-Id: I44f09aec60b0b18c458f9ba6d8b725db962e9cc7
    Closes-Bug: #1805164
    (cherry picked from commit 95cbc35b0bdae0de15569bc35fadbcf970da5b9e)
    (cherry picked from commit 315d9fab7aaa6b9835e80f5f56b49b44d382ca1f)

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/nova 19.0.0.0rc1 release candidate.

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/nova 17.0.10 release.

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/nova 18.2.0 release.