3.9.7.3 Modify an ACFS File System on Exascale

Exascale contains integrated support for Oracle Advanced Cluster File System (ACFS) on Exascale block storage using Exascale Direct Volumes (EDV).

To modify an existing Exascale-managed ACFS file system:

  1. Use the ESCLI acfsctl command to deregister the file system.

    If you deregister an Exascale-managed ACFS file system, the file system is dismounted and the mount point directory is removed on every node in the Oracle Grid Infrastructure (GI) cluster. The operation also removes the ACFS file system registration in the GI cluster. However, the file system contents remain in the underlying Exascale volume.

    To deregister an Exascale-managed ACFS file system, use the acfsctl deregister command and specify the file system identifier. For example:

    @> acfsctl deregister acfs0001_2fb06f13cddd4a8d8d636d1f794046cb

    You can use the lsacfsfilesystem command to find the identifier for each Exascale-managed ACFS file system.

    You can also bypass errors in the deregistration process by adding the --force option.

  2. After deregistering the file system, you can optionally remove and re-create the EDV attachment. See Administer Volume Attachments.

    Re-creating the EDV attachment enables you to move the attachment to a different GI cluster, which effectively moves the file system when it is registered again.

  3. Use the ESCLI acfsctl command to register the file system.

    During file system registration, you must specify:

    • The identifier of the Exascale volume that stores the file system. You can use the lsacfsfilesystem command to find the volume identifier associated with each Exascale-managed ACFS file system.

      The specified volume must be associated with an existing cluster-wide EDV attachment and contain existing ACFS file system, otherwise the command fails. Registration of a volume with a node-specific EDV attachment is not permitted.

      Furthermore, the ACFS file system is mounted on every node in the Oracle Grid Infrastructure (GI) cluster and the ACFS details are registered with the GI cluster. After registration, the file system is automatically mounted (or remounted) by Oracle Clusterware as required.

    • The mount point location where you want to mount the file system.

      The specified mount point location may differ from the location used in previous registrations.

    During file system registration, you may specify optional attributes governing how the file system is mounted. The permitted attributes are:

    • mountLeafOwner, mountLeafGroup, mountLeafMode: Specify the ownership and permission attributes applied to the leaf directory in the file system mount path.

    • mountNoRootSUID: Controls whether non-root users are permitted to run root-owned binaries with the set user ID (setuid) permission.

      • true: Only the root user can run root-owned setuid binaries.

      • false: Non-root users can also run root-owned setuid binaries.

      If not specified, the default value is false.

    • mountReadOnly: Specifies whether the file system is mounted with read-only access.

      • true: Mounts the file system with read-only access.

      • false: Mounts the file system with read-write access.

      If not specified, the default value is false.

    You cannot set any other file system attributes during registration.

    You can use the describe acfsfilesystems to see further details about all Exascale-managed ACFS file system attributes. See also Describing Resources and Attributes.

    For example:

    @> acfsctl register vol0001_bbd6fb4c75e2411b9bf366fe702eabaf /mnt/acfs1 --attributes mountReadOnly=true

    In the example:

    • The volume identifier is vol0001_bbd6fb4c75e2411b9bf366fe702eabaf.

    • The file system mount point is /mnt/acfs1.

    • The file system is mounted in read-only mode by specifying the optional attribute setting: mountReadOnly=true.