Cannot remove host from aggregate if host has been deleted

[Qin Zhao]

Kilo code

Reproduce steps:

1. Assuming that we have one nova-compute node named 'icm' which is added into one aggregate named 'zhaoqin'

[root@icm ~]# nova aggregate-details zhaoqin
+----+---------+-------------------+-------+--------------------------------+
| Id | Name | Availability Zone | Hosts | Metadata |
+----+---------+-------------------+-------+--------------------------------+
| 1 | zhaoqin | zhaoqin-az | 'icm' | 'availability_zone=zhaoqin-az' |
+----+---------+-------------------+-------+--------------------------------+
[root@icm ~]# nova service-list
+----+------------------+------+------------+---------+-------+----------------------------+-----------------+
| Id | Binary | Host | Zone | Status | State | Updated_at | Disabled Reason |
+----+------------------+------+------------+---------+-------+----------------------------+-----------------+
| 1 | nova-conductor | icm | internal | enabled | up | 2015-06-30T14:04:25.828383 | - |
| 3 | nova-scheduler | icm | internal | enabled | up | 2015-06-30T14:04:24.525474 | - |
| 4 | nova-consoleauth | icm | internal | enabled | up | 2015-06-30T14:04:24.640657 | - |
| 5 | nova-compute | icm | zhaoqin-az | enabled | up | 2015-06-30T14:04:19.865857 | - |
| 6 | nova-cert | icm | internal | enabled | up | 2015-06-30T14:04:25.080046 | - |
+----+------------------+------+------------+---------+-------+----------------------------+-----------------+

2. Remove the nova-compute using service-delete command. However, the host is still in aggregate.

[root@icm ~]# nova service-delete 5
[root@icm ~]# nova service-list
+----+------------------+------+----------+---------+-------+----------------------------+-----------------+
| Id | Binary | Host | Zone | Status | State | Updated_at | Disabled Reason |
+----+------------------+------+----------+---------+-------+----------------------------+-----------------+
| 1 | nova-conductor | icm | internal | enabled | up | 2015-06-30T14:05:35.826699 | - |
| 3 | nova-scheduler | icm | internal | enabled | up | 2015-06-30T14:05:34.524507 | - |
| 4 | nova-consoleauth | icm | internal | enabled | up | 2015-06-30T14:05:34.638234 | - |
| 6 | nova-cert | icm | internal | enabled | up | 2015-06-30T14:05:35.092009 | - |
+----+------------------+------+----------+---------+-------+----------------------------+-----------------+
[root@icm ~]# nova aggregate-details zhaoqin
+----+---------+-------------------+-------+--------------------------------+
| Id | Name | Availability Zone | Hosts | Metadata |
+----+---------+-------------------+-------+--------------------------------+
| 1 | zhaoqin | zhaoqin-az | 'icm' | 'availability_zone=zhaoqin-az' |
+----+---------+-------------------+-------+--------------------------------+

3. Then, attempt to remove the host from aggregate, but fails. And we can not remove this aggregate either, because it is not empty.

[root@icm ~]# nova aggregate-remove-host zhaoqin icm
ERROR (NotFound): Cannot remove host icm in aggregate 1: not found (HTTP 404) (Request-ID: req-b5024dbf-156a-44ee-b48e-fc53a331e05d)
[root@icm ~]# nova aggregate-delete zhaoqin
ERROR (BadRequest): Cannot remove host from aggregate 1. Reason: Host aggregate is not empty. (HTTP 400) (Request-ID: req-a3c5346c-9a96-49f4-a76d-a7baa768a0ef)

[OpenStack Infra (hudson-openstack)]

Related fix proposed to branch: master
Review: https://review.openstack.org/262165

[OpenStack Infra (hudson-openstack)]

Change abandoned by Michael Still (<email address hidden>) on branch: master
Review: https://review.openstack.org/262165
Reason: As mentioned, I am abandoning this spec to clear out the review queue ready for Newton. That said, feel free to restore this change with an updated version when you're ready to proceed.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/306192

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/306192
Committed: https://git.openstack.org/cgit/openstack/nova/commit/?id=1abac25fd2967be00c4d584c06ba15ca4e8d04cc
Submitter: Jenkins
Branch: master

commit 1abac25fd2967be00c4d584c06ba15ca4e8d04cc
Author: Danil Akhmetov <email address hidden>
Date: Fri Apr 29 11:45:28 2016 +0300

    Remove compute host from all host aggregates when compute service is deleted

    Nova currently does not check if compute host included in host-aggregates
    when user deletes compute service. It leads to inconsistency in nova host
    aggregates, impossibility to remove compute host from host-aggregate or
    remove host aggregate with invalid compute host.

    Change-Id: I8034da3827e47f3cd575e1f6ddf0e4be2f7dfecd
    Closes-Bug: #1470341

[OpenStack Infra (hudson-openstack)]

Change abandoned by Michael Still (<email address hidden>) on branch: master
Review: https://review.openstack.org/197648
Reason: This code hasn't been updated in a long time, and is in merge conflict. I am going to abandon this review, but feel free to restore it if you're still working on this.

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/nova 14.0.0.0b3 development milestone.