Chapter 39. File Systems

The CephFS kernel client is now available

Starting with Red Hat Enterprise Linux 7.3, the Ceph File System (CephFS) kernel module enables, as a Technology Preview, Red Hat Enterprise Linux nodes to mount Ceph File Systems from Red Hat Ceph Storage clusters. The kernel client in Red Hat Enterprise Linux is a more efficient alternative to the Filesystem in Userspace (FUSE) client included with Red hat Ceph Storage. Note that the kernel client currently lacks support for CephFS quotas. For more information, see the Ceph File System Guide for Red Hat Ceph Storage 2: https://access.redhat.com/documentation/en/red-hat-ceph-storage/2/single/ceph-file-system-guide-technology-preview (BZ#1205497)

ext4 and XFS file systems now support DAX

Starting with Red Hat Enterprise Linux 7.3, Direct Access (DAX) provides, as a Technology Preview, a means for an application to directly map persistent memory into its address space. To use DAX, a system must have some form of persistent memory available, usually in the form of one or more Non-Volatile Dual In-line Memory Modules (NVDIMMs), and a file system that supports DAX must be created on the NVDIMM(s). Also, the file system must be mounted with the dax mount option. Then, an mmap of a file on the dax-mounted file system results in a direct mapping of storage into the application's address space.
For a list of supported NVDIMM products and configurations, see the New kernel subsystem: libnvdimm entry in the Storage chapter, Part I. New Features. (BZ#1274459)

pNFS Block Layout Support

As a Technology Preview, the upstream code has been backported to the Red Hat Enterprise Linux client to provide pNFS block layout support.
In addition, Red Hat Enterprise Linux 7.3 includes the Technology Preview of the pNFS SCSI layout. This feature is similar to pNFS block layout support, but limited only to SCSI devices, so it is easier to use. Therefore, Red Hat recommends the evaluation of the pNFS SCSI layout rather than the pNFS block layout for most use cases. (BZ#1111712)

OverlayFS

OverlayFS is a type of union file system. It allows the user to overlay one file system on top of another. Changes are recorded in the upper file system, while the lower file system remains unmodified. This allows multiple users to share a file-system image, such as a container or a DVD-ROM, where the base image is on read-only media. Refer to the kernel file Documentation/filesystems/overlayfs.txt for additional information.
OverlayFS remains a Technology Preview in Red Hat Enterprise Linux 7.3 under most circumstances. As such, the kernel will log warnings when this technology is activated.
Full support is available for OverlayFS when used with Docker under the following restrictions:
  • OverlayFS is only supported for use as a Docker graph driver. Its use can only be supported for container COW content, not for persistent storage. Any persistent storage must be placed on non-OverlayFS volumes to be supported. Only default Docker configuration can be used; that is, one level of overlay, one lowerdir, and both lower and upper levels are on the same file system.
  • Only XFS is currently supported for use as a lower layer file system.
  • SELinux must be enabled and in enforcing mode on the physical machine, but must be disabled in the container when performing container separation; that is, /etc/sysconfig/docker must not contain --selinux-enabled. SELinux support for OverlayFS is being worked on upstream, and is expected in a future release.
  • The OverlayFS kernel ABI and userspace behavior are not considered stable, and may see changes in future updates.
  • In order to make the yum and rpm utilities work properly inside the container, the user should be using the yum-plugin-ovl packages.
Note that OverlayFS provides a restricted set of the POSIX standards. Test your application thoroughly before deploying it with OverlayFS.
Note that XFS file systems must be created with the -n ftype=1 option enabled for use as an overlay. With the rootfs and any file systems created during system installation, set the --mkfsoptions=-n ftype=1 parameters in the Anaconda kickstart. When creating a new file system after the installation, run the # mkfs -t xfs -n ftype=1 /PATH/TO/DEVICE command. To determine whether an existing file system is eligible for use as an overlay, run the # xfs_info /PATH/TO/DEVICE | grep ftype command to see if the ftype=1 option is enabled.
There are also several known issues associated with OverlayFS as of Red Hat Enterprise Linux 7.3 release. For details, see Non-standard behavior in the Documentation/filesystems/overlayfs.txt file. (BZ#1206277)

Support for NFSv4 clients with flexible file layout

Support for flexible file layout on NFSv4 clients was first introduced in Red Hat Enterprise Linux 7.2 as a Technology Preview. This technology enables advanced features such as non-disruptive file mobility and client-side mirroring, which provides enhanced usability in areas such as databases, big data and virtualization. This feature has been updated in Red Hat Enterprise Linux 7.3, and it is still offered as a Technology Preview.
See https://datatracker.ietf.org/doc/draft-ietf-nfsv4-flex-files/ for detailed information about NFS flexible file layout. (BZ#1217590)

Btrfs file system

The Btrfs (B-Tree) file system is supported as a Technology Preview in Red Hat Enterprise Linux 7.3. This file system offers advanced management, reliability, and scalability features. It enables users to create snapshots, it enables compression and integrated device management. (BZ#1205873)

pNFS SCSI layouts client and server support is now provided

Client and server support for parallel NFS (pNFS) SCSI layouts is provided as a Technology Preview starting with Red Hat Enterprise Linux 7.3. Building on the work of block layouts, the pNFS layout is defined across SCSI devices and contains sequential series of fixed-size blocks as logical units that must be capable of supporting SCSI persistent reservations. The Logical Unit (LU) devices are identified by their SCSI device identification, and fencing is handled through the assignment of reservations. (BZ#1305092)