Migrating an offline cluster to a new major version

When migrating from Informix 11.70 or a later version to 15.0.0, you stop all servers in the cluster during a period when downtime is acceptable. While the cluster if offline, you upgrade all the servers. You then start the primary followed by the secondary servers. You need not rebuild the cluster or clone the servers.

Prerequisites:

Be sure to stop and start the servers in the cluster in the order shown in the following procedure.

To migrate to a new version with high-availability clusters:

  1. Stop the Connection Manager by issuing the oncmsm -k connection_manager_name command.
  2. If you are using a High-availability Data Replication (HDR) secondary server as a backup server in case of contingencies:
    1. Quiesce the primary server by issuing an onmode -sy command to prevent user connections to the server.
    2. Force a checkpoint by issuing an onmode –c command on the primary server.
      Important: Do not proceed to the next step until that checkpoint has been processed on all secondary servers.
  3. Stop the secondary servers in the cluster in the following order:
    1. If you have remote standalone (RS) secondary servers, stop them by issuing the onmode -ky command.
    2. If you have shared disk (SD) servers, stop them by issuing the onmode -ky command.
    3. If you have an HDR secondary server, stop it by issuing the onmode -ky command.
  4. Stop the primary server by issuing the onmode -ky command.
  5. On each server, set the INFORMIXDIR environment variable to the full path name for the target installation.
  6. Ensure that all of the necessary configuration files are available in the target installation.
  7. Optional: Enable quick reversion to a consistent restore point if the migration fails. Do this by setting the CONVERSION_GUARD and RESTORE_POINT_DIR configuration parameters. (For more information, see Configuring for recovery of restore point data in case an upgrade fails.)
  8. Start the primary server by issuing an oninit command.
  9. Ensure that conversion of the primary was completed successfully and that the server is in multi-user mode.
  10. For HDR and RSS servers,
    1. Enable quick reversion to a consistent restore point if the migration fails. Do this by setting the CONVERSION_GUARD and RESTORE_POINT_DIR configuration parameters. (For more information, see Configuring for recovery of restore point data in case an upgrade fails.)
    2. Start secondary server by issuing an oninit command.
    3. Check ‘onstat -g cluster’ output at primary server, and once HDR/RSS secondary server connected to primary server, force a checkpoint at primary server using ‘onmode -c’ command and make sure checkpoint is replayed at HDR/RSS secondary server.
  11. Start the Connection Manager by issuing an oncmsm command.
  12. Start SD secondary servers by issuing oninit commands.
  13. Perform any additional standard migration tasks described in Migrating to the new version of Informix and in Completing required post-migration tasks.
  14. If you have Connection Manager configuration files that you created with a version of Informix Client Software Development Kit (Client SDK) that is prior to version 3.70.xC3, you must convert the files. You must convert them because the older files are incompatible with the current version of the Connection Manager. For more information, see Converting older formats of the Connection Manager configuration file to the current format.

The in-place migration of all servers in the cluster is now complete. They are now running in what is called compatibility mode. Legacy applications may be tested in this mode and should work as they did in the previous version. Reversion of the primary server, if it becomes necessary, is allowed while in compatibility mode. New applications may also be tested in compatibility mode. Be aware, however, that many 15.0.0 features are not enabled.

In order to enable all version 15.0.0 features one must move the primary server out of compatibility mode and into native mode.

Note:

Reversion is not supported for HDR and RSS secondary servers. To revert a cluster you must shut down all secondary servers, revert the primary server, and then rebuild the cluster by recreating all secondary servers. You can use the ifxclone utility to recreate secondary servers.