After an uncontrolled failover, for Oracle SI and RAC complete the following procedure:

On the target environments:

  1. Manually shutdown all VDBs, all vCDBs and any CDBs containing vPDBs using sqlplus command SHUTDOWN ABORT.

  2. Kill all the stale Oracle processes of the database instances that were shutdown in above step.

    1. Run ps -ef | grep <ORACLE_SID> to get all the running processes associated with the database instance.

      For example ps -ef | grep VDBOMSR8A1718_FPX.

    2. Run kill -9 <process id1> <process id2> ... to kill all the running processes.

      For example: kill -9 15236.

  3. Remove all the Delphix mounts of the Oracle virtual sources from the target environment.

    1. List all the mounts on the target environment using the mount | grep <source-engine-ip> command.

    2. Run umount -lf <Delphix-mount> to remove the mounts. This might require sudo privileges.

      For example: sudo umount -lf /mnt/provision/VDBOMSR8A1718_FPX

      Use the appropriate syntax for umount for your operating system. The above example is for Linux.

  4. Manually start linked CDBs using sqlplus (or srvctl for RAC database) .

On the target Delphix Engine:

  1. Enable all the source environments and then the target environments.

  2. Enable all the Oracle non-multitenant and PDB dSources.

  3. Enable vPDBs in the linked CDBs.

    If a vPDB was enabled on the source at the time of the last replication update before the outage, enabling it here may fail with an Oracle error indicating a datafile requires media recovery (for example, ORA-01113). If this occurs, a normal Enable, Refresh, or Rewind cannot bring the vPDB online — a forced Rewind (or Refresh) must be performed via the CLI instead, using the force parameter of OracleRollbackParameters/OracleRefreshParameters. See Refresh or Rewind Broken or Unusable Virtual PDBs and CLI Cookbook: Force Refresh or Rewind Virtual PDBs for details. A forced Rewind/Refresh cannot be undone with Undo Refresh/Rewind, so use it only when a normal Enable fails this way.

  4. Enable VDBs.

    Enable operation may fail due to an existing lock file for the database. Check the Oracle alert log file of the database for more details and the lock file location. Remove the lock file if required, after making sure that there are no database processes running. Then enable the VDB again.

  5. Enable virtual CDBs. Then enable any vPDBs in it.

Last updated
Oct 9, 2026