Wallaby HA overcloud deployed with bad pacemaker container image names
Affects | Status | Importance | Assigned to | Milestone | |
---|---|---|---|---|---|
tripleo |
In Progress
|
Undecided
|
Damien Ciabrini |
Bug Description
With the move to ephemeral Heat, the deployment of a HA overcloud no longer include an environment file that configures Heat parameter ClusterFullTag. Consequently, the control plane is deployed with pacemaker resource configured differently than what was expected to be the new default in Wallaby:
* Container bundle: haproxy-bundle [104.130.
* Container bundle: galera-bundle [104.130.
* Container bundle: rabbitmq-bundle [104.130.
* Container bundle: openstack-
The expected default should be akin to:
* Container bundle set: haproxy-bundle [cluster.
* Container bundle set: galera-bundle [cluster.
* Container bundle set: rabbitmq-bundle [cluster.
* Container bundle: openstack-
As a side effect, one can not run an minor update from e.g. Wallaby to Xena without causing a service disruption, because any change of container image name makes pacemaker restart the service globally.
Changed in tripleo: | |
status: | Confirmed → In Progress |
Related fix proposed to branch: stable/wallaby /review. opendev. org/c/openstack /tripleo- ansible/ +/861610
Review: https:/