Migrating an existing IBM Storage Scale cluster management to the IBM Storage Scale native REST API
You can migrate an existing IBM Storage Scale cluster from mm-commands CLI-based management to the IBM Storage Scale native REST API.
- Create a new cluster by using the native REST API through the scalectl CLI or REST API.
- Migrate each of the existing classic IBM Storage Scale cluster by using the native REST API migration tool.
- You can do the native REST API cluster migration as an online operation to a production cluster. However, as a best practice, run the migration within a maintenance window to allow for post-migration testing before full production use.
- Install the native REST API RPM on each of the existing classic IBM Storage Scale cluster nodes before running the cluster migration. The native REST API RPM installation migrates the directory structures and files that are owned by the IBM Storage Scale software. These updates set a new specific native REST API user and group where needed. Specifically, the /var/mmfs directory is modified for native non-root scaleadmd server access.
- Create a native REST API nodeId ahead of time for each existing classic IBM Storage Scale cluster node. This requirement is similar to requiring passwordless SSH keys to be configured with the classic IBM Storage Scale setup.
- Make sure that the all nodes in the classic IBM Storage Scale cluster, regardless of its starting software version, have the same native REST API generally available (GA) level of code (RPMs) installed. Also, set the minReleaseLevel to the GA level that introduces the native REST API (5.2.3.0) or later.
- Make sure that all existing IBM Storage Scale file systems in the classic IBM Storage Scale cluster are at the same GA level that introduces the native REST API (5.2.3.0) or later.
Use the optional --precheck option to validate each node and the cluster requirements before making any updates during the migration. It is recommended that you run the --precheck option before the full migration to identify potential issues early. If any issues arise during the --precheck, address them and rerun the --precheck until no further issues are identified. Then, run the full migration without supplying the --precheck option.
Since all nodes in a cluster are required to pass individual migration before the cluster will be committed for use with native REST API management, some node failures might occur. Error information is reported per node or at the cluster level if this occurs. Once the concerns are addressed on the failing nodes or at the cluster level, you can re-run the node migration.
For more information about the scalectl cluster migrate command, see scalectl cluster command.
For more information, see Coexistence with mm-commands CLI.