Restoring to a previous consistent state after a failed upgrade

If the CONVERSION_GUARD configuration parameter is enabled and an upgrade fails, you can run the onrestorept command to undo the changes that are made during the upgrade and restore OneDB to a consistent state.

Before you begin

The following prerequisites must be met:

  • The directory that is specified by the RESTORE_POINT_DIR configuration parameter contains the files that were stored during the failed upgrade attempt.
  • The server must be offline before you run the onrestorept -y command.

About this task

To restore the server to a previous consistent state after a failed upgrade:

Procedure

Run the onrestorept -y command, which displays prompts while the command runs. If you do not specify -y, you must respond to every prompt that appears.
Important: Do not start the server until the onrestorept utility finishes running. Starting the server before the onrestorept utility finishes running can damage the server, requiring the server to be restored from a backup copy.

What to do next

After you restore the server to a consistent state, you can resume work in the source version of the server or find and fix the problem that caused the upgrade to fail.
Important: To start Enterprise Replication after the onrestorept utility restores the database server to a consistent state, you must use the cdr cleanstart command.

Before you attempt another upgrade, run the onrestorept -c command to remove the files in the directory that is specified in the RESTORE_POINT_DIR configuration parameter. After a successful upgrade, the server automatically deletes restore point files from that directory.