PrerequisiteMigrating nodes require Cluster B to be online and both tdqg-manager and tdqg-node on Cluster B to be running QueryGrid version 02.08.00.00 and later. Also, the nodes must be able to access at least one the QueryGrid Managers in Cluster A. If these conditions are not met, the nodes cannot be migrated and an option to cancel the migration becomes available. The nodes can be re-added using the method used to initially add the node to the system.
Use the following procedure, with the terms Cluster A and Cluster B as examples, to merge two clusters.
If merging a site with a full Vantage stack and analytical nodes, open an incident with the QueryGrid GSO team for consultation before merging.
- Create a Backup of one of the QueryGrid Managers in Cluster B to merge.
- Copy the newly created backup from the Cluster B QueryGrid Manager to one of the Cluster A QueryGrid Managers.
- Run the migrate command as the root or tdqgm user on the Cluster A QueryGrid Manager where you copied the Cluster B backup file and answer the prompts./opt/teradata/tdqgm/bin/migrate.sh backup-file
Where backup_file is the path name of the backup ZIP file.
- Review the summary of actions to be performed before selecting Y or N to continue.Selecting N aborts the migration.
- In the QueryGrid portlet, validate that Cluster A contains the configuration from both Cluster A and Cluster B.When the migration is successful, nodes previously registered to Cluster B now display as owned by Cluster A and offline on Cluster B.
- If the migration fails, view details of the migration in the /var/opt/teradata/tdqgm/logs/tdqgm-command.log file.
- Run the /opt/teradata/tdqgm/bin/create-join-token.sh command on one of the QueryGrid Managers from Cluster A and record the following information from the output:
Command Description join host The public address of the QueryGrid Manager used by nodes to connect to the QueryGrid Manager. join token The temporary authentication token enabling other QueryGrid Managers to join the cluster. cluster ID The UUID identifying the cluster. - Join the QueryGrid Managers from Cluster B to Cluster A, one at a time, using the /opt/teradata/tdqgm/bin/join-cluster.sh command on each QueryGrid Manager from Cluster B.
- If the same instance of Viewpoint is used to monitor both clusters, delete Cluster B from the Monitored Systems portlet; otherwise, re-add the certificate on the Cluster B Viewpoint instance so the QueryGrid portlet continues to work for the new cluster.