Openstack uses the placement-service for tracking internal resources. From Newton it was an integrated part of Nova, and from Stein it can be deployed as its own service. From train it have to be its own service, so we need to install the stand-alone placement service before we can upgrade to train. This article will guide you through this process
| Note |
|---|
Make sure to have recent enough versions of our puppet-repositories. |
Install placement API servers
...
Create a couple of infra-VM's and install the puppet-role role::openstack::placement
...
Prepare the migration of data
Download the migration-script, and create a config-file for it:
| Code Block |
|---|
# wget https://opendev.org/openstack/placement/raw/branch/master/placement_db_tools/mysql-migrate-db.sh
# ./mysql-migrate-db.sh --mkconfig migrateconfig.tc
Wrote /root/migrateconfig.tc |
Populate the config-file with relevant values. Install mariadb-client.
Migrate data from nova-placement to the new placement service
Add hiera-key to point haproxy to new placement-service
...
The following list helps you to actually perform the migration. This procedure WILL make downtime for your nova-API's, so it have to be done in a maintainance-period.
- Stop puppet and apache2 on all novaapi-servers
- Run the migration
# ./mysql-migrate-db.sh --migrate migrateconfig.tc - Set the key in hiera which tells our infrastructure that the new placement-service should be used, and make sure to give puppet a run on the adminlb's.
ntnuopenstack::placement::standalone: true
- Re-start
...
- apache2 on novaapi-servers.