Upgrade paths and commands for file audit logging and clustered watch folder

Use the following information to upgrade file audit logging and clustered watch folder in IBM Storage Scale.

File audit logging base functionality was introduced in IBM Storage Scale 5.0.0.

Clustered watch folder base functionality was introduced in IBM Storage Scale 5.0.3.

In IBM Storage Scale 5.1.0, there is a large architecture change that removes the dependency for the message queue and the gpfs.kafka package. The mmmsgqueue config --remove-msgqueue command automatically upgrades all components that are needed and disables then re-enables all of the existing healthy audits and watches. When you run this command, there is a small outage where no events are generated. The prerequisites to run this command are as follows:

  • Run mmchconfig release=LATEST first.
  • All file system versions where file audit logging or clustered watch folder are enabled or are planned to be enabled must set to at least IBM Storage Scale 5.1.0 or later. See mmlsfs <fs> -V to query the file system version.
  • When upgrading file audit logging and clustered watch folder in remote cluster environments, you must ensure the following:
    • For file audit logging, ensure that your system meets the requirements for using file audit logging with remotely mounted file systems. For more information, see Requirements for using file audit logging with remotely mounted file systems.
    • To avoid version conflicts, it is required that you upgrade the accessing clusters before upgrading the owning cluster. Unexpected errors might occur on the remotely mounted file systems with file audit logging enabled if the remote cluster is at IBM Storage Scale 5.0.1 or earlier and the owning cluster was upgraded first.
Note: For more information about issues that might occur during an upgrade, see File audit logging issues.

If file audit logging or clustered watch folder has never been enabled or your cluster was installed at 5.1.2 or later, the mmmsgqueue config --remove-config command does not need to run.

If you are upgrading from 5.0.5.X with msgqueue enabled, you must disable file audit logging and clustered watch folder and then run mmmsgqueue config --remove before the upgrade. File audit logging and clustered watch folder can be enabled after the upgrade completes.

If you are upgrading from 5.1.0.X/5.1.1.X with msgqueue enabled, you must run mmmsgqueue config --remove-msgqueue prior to the upgrade. This re-enables file audit logging and clustered watch folders automatically.

After a full cluster upgrade to 5.1.3 and above, regardless of file system level, the mmrestorefs command will not restore files that are located in the .audit_log fileset or configuration fileset. The current configuration will not be overwritten and audit records will not be removed or restored.

If running file audit logging or watch folder in a mixed release environment, issue the management commands from the lowest release level node.