Getting ServiceTooOld after an offline upgrade

[Gorka Eguileor]

On offline upgrades, due to the rolling upgrade mechanism we need to restart the cinder services twice to complete the upgrade, just like in the rolling upgrade case, otherwise we'll see ServiceTooOld exceptions on the new API calls.

The current offline upgrade process is:

- Stop cinder services
- Upgrade the cinder nodes
- Sync your DB
- Start the cinder services
- Restart all the cinder services

This second restart creates a bad user experience and it should not be necessary for offline upgrades.

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: master
Review: https://review.openstack.org/553772

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/553772
Committed: https://git.openstack.org/cgit/openstack/cinder/commit/?id=3cd2ebd3759c76fdf5a292e612127094c7aa2b17
Submitter: Zuul
Branch: master

commit 3cd2ebd3759c76fdf5a292e612127094c7aa2b17
Author: Gorka Eguileor <email address hidden>
Date: Fri Mar 16 13:11:59 2018 +0100

    Avoid second restart on offline upgrades

    On offline upgrades, due to the rolling upgrade mechanism we need to
    restart the cinder services twice to complete the upgrade, just like in
    the rolling upgrade case.

    The current offline upgrade process is:

    - Stop cinder services
    - Upgrade the cinder nodes
    - Sync your DB
    - Start the cinder services
    - Restart all the cinder services

    This second restart creates a bad user experience and it should not be
    necessary on an offline upgrade, so this patch adds a new optional
    parameter -called "--bump-versions"- to the cinder-manage db sync
    command that allows us to skip the restart of the services.

    Closes-Bug: #1756321
    Change-Id: I1b58c637f6b2187a78c9c00a6c4933335439ad6f

[OpenStack Infra (hudson-openstack)]

Fix proposed to branch: stable/queens
Review: https://review.openstack.org/560335

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/cinder 13.0.0.0b1 development milestone.

[OpenStack Infra (hudson-openstack)]

Reviewed: https://review.openstack.org/560335
Committed: https://git.openstack.org/cgit/openstack/cinder/commit/?id=25c737d6b8e19a1932696554e47dd262ae651592
Submitter: Zuul
Branch: stable/queens

commit 25c737d6b8e19a1932696554e47dd262ae651592
Author: Gorka Eguileor <email address hidden>
Date: Fri Mar 16 13:11:59 2018 +0100

    Avoid second restart on offline upgrades

    On offline upgrades, due to the rolling upgrade mechanism we need to
    restart the cinder services twice to complete the upgrade, just like in
    the rolling upgrade case.

    The current offline upgrade process is:

    - Stop cinder services
    - Upgrade the cinder nodes
    - Sync your DB
    - Start the cinder services
    - Restart all the cinder services

    This second restart creates a bad user experience and it should not be
    necessary on an offline upgrade, so this patch adds a new optional
    parameter -called "--bump-versions"- to the cinder-manage db sync
    command that allows us to skip the restart of the services.

    Closes-Bug: #1756321
    Change-Id: I1b58c637f6b2187a78c9c00a6c4933335439ad6f
    (cherry picked from commit 3cd2ebd3759c76fdf5a292e612127094c7aa2b17)

[OpenStack Infra (hudson-openstack)]

This issue was fixed in the openstack/cinder 12.0.2 release.