Backup and recovery considerations for CIMOM

It is important to schedule backups of the repository directories and files. If the repository is moved, is lost, or becomes corrupted, restore the files that you have backed up.

In 5770-UME V1R4M0, command line utility “cimreparchive” is shipped as an additional mechanism to backup repository. Please refer to cimreparchive usage information for details.

Here are the namespaces that are installed with CIMOM:

root
The root namespace conforms to the Distributed Management Task Force (DMTF) specifications.
root/cimv2
This is for standard CIM schemas for the shipped providers.
root/PG_InterOp
This is for provider registration. This space is reserved exclusively for providers and all providers must be registered.
root/PG_Internal
This space is reserved and used by CIMOM.
root/ibmsd
The namespace is owned and used by IBM Systems Director.

Backup

It is recommended that the CIM server UserData/ directory be backed up. Files in UserData are not expected to change very often in normal cases. They are mainly files that are created during set up, and are not changed. However, since applications or providers might change these files more often, a daily back up would be recommended. User may add content into repository or other files into UserData, the information will be mixed with IBM shipped information.

Note: The documentation requires distinguish between IBM files and User files in the UserData/ directory, for it is not seem to be worth the same effort. The whole UserData/ directory should be backed up. The customer could run the backup command to only save the changed files since the last backup.

Notes® about specific files and directories:

  • The repository as a whole can be backed up, including any temporary transaction files.
  • The server configuration file (current and planned) can be backed up.
  • The SSL files in the UserData/ directory are not considered IBM files. These files can be backed up, including the server certificate, the private key, and trust stores. However, the backup location should be secure.
  • The mappings files in the UserData/ directory are not considered IBM files. These files can be backed up.
  • The migration marker file is a user file that can be backed up.
  • The trace files and log files are user files that can be backed up.

The Universal Manageability Enablement program has product files in its Proddata directory and includes the QUME product library, which would follow the recommended weekly backup. These are all “IBM” files that are installed with the UME licensed program, or PTF’d. All of these files can be restored from the backup media without special considerations.

Backup can be done by system command SAV, SAVLIB, some examples:
  • CRTSAVF FILE(TEST) and SAV DEV('/QSYS.lib/QGPL.LIB/TEST.FILE') OBJ(('/QOpenSys/QIBM/UserData/UME/Pegasus'))
  • CRTSAVF FILE(TESTLIB) and SAVLIB LIB(QUME) DEV(*SAVF) SAVF(TESTLIB)

Recovery

There are files in the CIM server’s UserData/ directory which are shipped as IBM data.

System management applications (Such as IBM Systems Director) may add User files to the CIM server’s UserData/ directory as part of their setup. The shipped IBM providers may add “User” files when they are called by the CIM server.

Finally, since the interface to change the CIM server’s UserData/ directory will be published, customers might want to add their own User data to this directory. Customers may configure the server, add SSL certificates, update the repository, etc.

Therefore, it is possible for the CIM server’s UserData/ directory to have a combination of IBM data and “User” data. In the CIM server design, the IBM data and “User” data are intermingled as files in the same directory, and sometimes as data in the same file.

The files that could contain both IBM and User data are the following:
  1. All instance index files and instance data files. The most likely case of intermingled IBM and User data instances would be the provider registrations. Any providers developed by customers would be registered in the same files as the IBM shipped providers.
  2. Instance association and class association files. These files keep track of associated classes or instances. Any associations created by customers would be stored in the same file as the IBM shipped associations.
  3. Server configuration files.

The files which may contain both IBM and User data present a problem for recovery. A process that simply restores the IBM shipped file from the install media would recover the IBM data but would lose the User data. However, the goal is to recover the IBM data without destroying customer data. So to summarize: If you need to recover the ProdData/ directory of the CIM server, reinstall the QUME product library. If you need to recover the IBM data in the UserData/ directory of the CIM server, reinstall the QUME library or recover these files from the backup media. If you need to recover the user data in the UserData/ directory of the CIM server, recover the destroyed files from the backup media.