sc000 upgrade job goes N -> N instead of N-1 -> N

Bug #1786488 reported by Jiří Stránský
6
This bug affects 1 person
Affects Status Importance Assigned to Milestone
tripleo
Fix Released
Critical
Sofer Athlan-Guyot

Bug Description

It seems we lost correct handling of overcloud_release and/or overcloud_docker_registry_namespace in upgrade jobs. E.g. upgrade jobs on master don't deploy queens containers like they should, they already start on master.

E.g. http://logs.openstack.org/87/589487/2/check/tripleo-ci-centos-7-scenario000-multinode-oooq-container-upgrades/82b2d19/logs/undercloud/home/zuul/

In upgrade jobs, overcloud_prep_containers.log.txt.gz should always download 1 release older images than upgrade_overcloud_prep_containers.log.txt.gz, but currently it downloads the same release.

tags: added: alert
Revision history for this message
Sofer Athlan-Guyot (sofer-athlan-guyot) wrote :

hi,

so the issue is there http://logs.openstack.org/87/589487/2/check/tripleo-ci-centos-7-scenario000-multinode-oooq-container-upgrades/82b2d19/job-output.txt.gz#_2018-08-09_10_50_03_763013 ... we're skipping the task that set the proper value for deployment.

So either container_build_id is defined or use_overcloud_mixed_upgrade is false

Revision history for this message
Sofer Athlan-Guyot (sofer-athlan-guyot) wrote :

Wrong, just ignore my previous comment, the skip doesn't happen during deployment of the overcloud ... the real one is there [1] and it's not skipped.

[1] http://logs.openstack.org/87/589487/2/check/tripleo-ci-centos-7-scenario000-multinode-oooq-container-upgrades/82b2d19/job-output.txt.gz#_2018-08-09_11_26_43_972408

Revision history for this message
Sofer Athlan-Guyot (sofer-athlan-guyot) wrote :
Revision history for this message
Sofer Athlan-Guyot (sofer-athlan-guyot) wrote :

So the problem is that we don't use the new docker_registry_namespace_used value as according we use the containers-prepare-parameter.yaml[1] which is created during undercloud deployment[2] with the assumption that overcloud and undercloud match.

We have to recreate that file during mixed upgrade:
 - when we deploy overcloud;
 - when we upgrade

[1] https://github.com/openstack/tripleo-quickstart-extras/blob/master/roles/overcloud-prep-containers/templates/overcloud-prep-containers.sh.j2#L33..L48

[2] https://github.com/openstack/tripleo-quickstart-extras/blob/master/roles/undercloud-deploy/tasks/create-scripts.yml#L107..L115

Revision history for this message
Matt Young (halcyondude) wrote :

patch posted:

https://review.openstack.org/#/c/590912

Don't use the undercloud containers during mixed upgrade.

When doing mixed upgrade we don't want to use the container definition
file generated during undercloud deployment.

The problem is that at that time (undercloud deployment) the build_id
and docker_registry_namespace_used are (rightly) pointing to master.
But then during mixed upgrade overcloud deployment we need the N-1
version docker's images.

Change-Id: Ice13dceba7417e1ef827f4ab44fd08d2d13922fd
Closes-Bug: #1786488

Changed in tripleo:
assignee: nobody → Sofer Athlan-Guyot (sofer-athlan-guyot)
Revision history for this message
Matt Young (halcyondude) wrote :

unclear why LP isn't being updated from commit messages, sofer's patch isn't the first I've seen this happen to in the past day

Revision history for this message
OpenStack Infra (hudson-openstack) wrote : Fix proposed to tripleo-quickstart-extras (master)

Fix proposed to branch: master
Review: https://review.openstack.org/590912

Changed in tripleo:
status: Triaged → In Progress
Revision history for this message
OpenStack Infra (hudson-openstack) wrote : Fix merged to tripleo-quickstart-extras (master)

Reviewed: https://review.openstack.org/590912
Committed: https://git.openstack.org/cgit/openstack/tripleo-quickstart-extras/commit/?id=07d14c4f68f3541a55ed12b0121cc8b1a2f10e97
Submitter: Zuul
Branch: master

commit 07d14c4f68f3541a55ed12b0121cc8b1a2f10e97
Author: Sofer Athlan-Guyot <email address hidden>
Date: Fri Aug 10 18:41:50 2018 +0200

    Don't use the undercloud containers during mixed upgrade.

    When doing mixed upgrade we don't want to use the container definition
    file generated during undercloud deployment.

    The problem is that at that time (undercloud deployment) the build_id
    and docker_registry_namespace_used are (rightly) pointing to master.
    But then during mixed upgrade overcloud deployment we need the N-1
    version docker's images.

    Change-Id: Ice13dceba7417e1ef827f4ab44fd08d2d13922fd
    Closes-Bug: #1786488

Changed in tripleo:
status: In Progress → Fix Released
Revision history for this message
Jose Luis Franco (jfrancoa) wrote :
Changed in tripleo:
status: Fix Released → In Progress
Revision history for this message
Jose Luis Franco (jfrancoa) wrote :

Opening a new LP

Changed in tripleo:
status: In Progress → Fix Released
Revision history for this message
OpenStack Infra (hudson-openstack) wrote : Fix proposed to tripleo-quickstart-extras (master)

Fix proposed to branch: master
Review: https://review.openstack.org/595143

Revision history for this message
OpenStack Infra (hudson-openstack) wrote : Change abandoned on tripleo-quickstart-extras (master)

Change abandoned by Athlan-Guyot sofer (<email address hidden>) on branch: master
Review: https://review.openstack.org/595143
Reason: I don't remember if it's still need or not and the lp is kinda confusing. So let's abandon this.

Revision history for this message
OpenStack Infra (hudson-openstack) wrote : Fix included in openstack/tripleo-quickstart-extras 2.1.1

This issue was fixed in the openstack/tripleo-quickstart-extras 2.1.1 release.

To post a comment you must log in.
This report contains Public information  
Everyone can see this information.

Other bug subscribers

Remote bug watches

Bug watches keep track of this bug in other bug trackers.