Chapter 11. Storage

The installer now allows you to create a thinly-provisioned LVM (Logical Volume Management) layout. Support is limited to custom Kickstart installations only; it is not possible to create a LVM thin provisioning layout automatically using the autopart Kickstart command, and you cannot select this storage configuration during an interactive installation using the graphical or text-based user interface.

Additional mount points and a list of allowed mount options can now be specified with udev rules. The system administrator can write a custom rule to enforce or limit mount options for a specific set of devices. For example, USB drives can be limited to be always mounted as read-only.

The udisks tool now accepts the noexec global option to be enforced on all unprivileged users' mount points. On desktop systems, the noexec option can protect users from mistakenly running certain applications.

Previously, default settings for the Dell MD36xxf storage arrays were not included in the devices section of the default multipath configuration file, which affected performance for these arrays. These settings are now included in this configuration file.

Users were unable to split their configuration between /etc/multipath.conf and other configuration files. This prevented users from setting up one main configuration file for all their machines and keeping machine-specific configuration information in separate configuration files for each machine.

To address this, a new config_dir option was added in the multipath.config file. Users must change the config_dir option to either an empty string or a fully qualified directory path name. When set to anything other than an empty string, multipath will read all .conf files in alphabetical order. It will then apply the configurations exactly as if they had been added to the /etc/multipath.conf. If this change is not made, config_dir defaults to /etc/multipath/conf.d.

If a logical volume is read-only and active but its metadata states that it should be writeable (a situation that can arise if the configuration setting activation/read_only_volume_list is changed), you can now use the lvchange --permission rw command to bring the active copy in line with the metadata and make it writeable. Executing an lvchange --refresh command can do this as well, but this new feature might be more convenient in some circumstances. The opposite is also true: The lvchange --permission r command will now refresh an active logical volume that should be read-only. For information on the lvchange command, see the lvchange(8) man page.

For multipathd to stop attempting to use a path, that path must be inaccessible for the timeout period of 300 seconds. This can give the appearance that multipathd has stalled. Two configuration options have been added to improve user experience: delay_watch_checks and delay_wait_checks. Use delay_watch_checks to specify the number of cycles that multipathd should wait before using a path that has just come online. If the path fails in fewer cycles than the value specified, multipathd will not use the path. Then, use the delay_wait_checks parameter to specify the number of cycles that a path must work correctly until it can be considered accessible again. This prevents unreliable paths from being used immediately when they come back online.

The 3.3.2 version of mdadm provides a number of bug fixes as well as features such as automatically rebuilding an array in the event of a failed RAID volume, RAID level migrations, check-pointing fault tolerance, and SAS-SATA drive roaming. These features are supported on external metadata formats and continues Red Hat's support of Intel's RSTe SW stack.

The lvmconf script now provides --enable-halvm and --disable-halvm options to set proper configuration in the etc/lvm/lvm.conf configuration file that is suitable for an HA-LVM environment. In addition, the lvmconf script now also provides --service, --mirrorservice, and --startstopservices options to enable or disable related SysV init services needed for LVM to work correctly in a clustered environment. For information on the lvmconf script, see the lvmconf(8) man page.

The iprutils packages have been upgraded to upstream version 2.4.5, which provides a number of bug fixes and enhancements over the previous version. Notably, this update adds support for reporting cache hits on the Serial Attached SCSI (SAS) disk drive, and increases the speed of array creation for an advanced function (AF) direct-access storage device (DASD).

The dm-cache device-mapper target, which was previously included as a Technology Preview, is now fully supported. This device-mapper target allows fast storage devices to act as a cache for slower storage devices. See the lvmcache manual page for more information.