Execution steps
You must prepare all the services for transition.
Group transition steps
On the left navigation pane, the migration steps are grouped by phases awaiting execution. These phases are differentiated by their effect on the production cluster.
Ansible scripts are run on the control node. However, most tasks are run on the cluster nodes also. As a consequence, CMA requires Python and some packages (see <CMA_ROOT_DIR>/am2cm-ansible/requirements.txt) on all nodes to work.
- To start the upgrade process, click
. - To run multiple tasks in phases , click
. You can click Run All in Current Phase or
Run All. However, Cloudera recommends you
to use Run All on dev clusters only.
The CMA Python setup creates temporary Python 3.9.8 virtual environments on all nodes, unless there is already a higher version installed. The cluster nodes can be on an air gap environment. However, the control node needs access to the Internet.
You can run the Pre-transition Validation steps anytime and does not disturb the cluster. Performing the steps in this phase ensures that your cluster is ready for upgrade.
To take the backup, the cluster must be in the maintenance mode even if it does not change payload data. If you have have additional (non-HDP) services and application, then you need additional backup implementation in-place. You need to add a custom backup steps extending the cma-server/resources/transitions/am2cm/sections/hdpX-backup.yml
The Transition and Post-transition steps require a freshly installed empty Cloudera Manager with Cloudera Management Service that is compatible with the intended CDP runtime version. For information, see the Cloudera Manager Installation and Setup and Installing Cloudera Management Service documentation.
During the transition, there is a change in the binaries and update configurationss. However, in post-transition steps, the payload data structure is changed for compatibility with the new service versions.
If you have a private parcel repository location available, you can change it by clicking .
Individual transition steps
You can manually transition by clicking 
and using the documentation. After the manual
transition is complete, you must notify the CMA server by clicking 
to indicate that
the manual transition is complete.
However, the preferred approach is to use the Ansible script. The UI calls these scripts one by one based on their tags. You can change some of the input parameters for these scripts. Also, the scripts are part of CMA allowing you to modify the scripts. For information, see <CMA_ROOT_DIR>/am2cm_ansible. To modify the transition flow, you need to edit the .yml file that is available. For information, see folder <CMA_ROOT_DIR>/cma-server/config/transitions/am2cm.
For every service there is a tag displayed. You can use the tag to identify the ansible task being executed. The actual ansible command is called in the control node where the CMA server (cma-server) is located and is shown below the tag. After you click the icon, output and error log tabs appear below the ansible command, allowing you to check the output logs associated with the ansible upgrade playbook that is executed.
These scripts are manually executable on the am2cm-ansible directly on the control node (container) and also encouraged to be modified when needed.
