MariaDB does not officially supports downgrade between
major versions [1].
When the update fails and the application framework tries
to roll-back the STX-O deployment, the recovery fails
in the MariaDB Helm chart.
The solution for this is first, when the update fails it will
trigger the deletion of the updated MariaDB Helm release.
The other step will be to have the persistentvolume to the old
MariaDB backed up at the beggining of the update, and if the update
fails, it will trigger a function that will remove the
persistentVolume from the new MariaDB and replace with the old one.
this work is going to to be maped to the task 51826 in the storyboard:
https://storyboard.openstack.org/#!/story/2011262
This review aims to address the first part of the solution, where
the MariaDB Helmrelease is deleted before starting the recovery
process.
It was necessary to create a function do delete the Helmrelease as
it was noted that the new MariaDB could be redeployed by FluxCD
reconciliation, before the old one, if it was opted to only
uninstall the Helmrelease.
Test Plan:
PASS - Build python3-k8sapp-openstack
PASS - Build WRO
PASS - Force an update fail and check if MariaDB Helmrelease is deleted
Relates-To: https://review.opendev.org/c/starlingx/openstack-armada-app/+/945094/6
Both reviews on the relation chain should be merged together as it
functions as one to MariaDB be able to rollback in case of failed
update.
[1] - https://mariadb.com/kb/en/downgrading-between-major-versions-of-mariadb/
Story: 2011262
Task: 51787
Change-Id: I60b009c805673d00d54b1c4ff8c868cd8f2f0190
Signed-off-by: Daniel Caires <DanielMarques.Caires@windriver.com>
a3310458a3