1.163. mdadm

An updated mdadm package that fixes several bugs and adds various enhancements is now available for Red Hat Enterprise Linux 6.
mdadm is a utility for creating, managing, and monitoring Linux MD (multiple disk) devices.
Bug Fixes
BZ#605710
Previously, the "noiswmd" kernel command line option did not set the rd_NO_MDIMSM variable to 1 and the udev rules thus failed to match the option. This update removes the rule for the rd_NO_MDIMSM variable and adds the "noiswmd" and "nodmraid" command line options, which substitute the rule for rd_NO_MDIMSM.
BZ#626762
The md kernel module stopped responding when attempting to stop a RAID device. This occurred due to a mutex deadlock. With this update, the underlying code was changed and the issue no longer occurs.
BZ#636883
Previously, mdadm did not accept the short version of the "--export" (that is "-Y") and "--name" (that is "-N") options. This was due to missing entries in the list of short options. With this update, the missing entries have been added and the short versions of the options now work as expected.
Enhancements
BZ#633306
Previously, mdadm was not able to rebuild newly-connected drives automatically. This update adds the array auto-rebuild feature and allows a RAID stack to automatically rebuild newly-connected drives.
BZ#633667
RAID migration is now supported: mdadm is now able to change the RAID level of an already-existing device.
BZ#633671
Previously, mdadm could not add a new disk to an already-existing volume. This update adds the OLCE (On Line Capacity Expansion) feature, which allows mdadm to add new disks to an array and allocate their resources to existing volumes.
BZ#633688
Previously, mdadm did not track the progress of the rebuild and level-migration operations and therefore was not able to recover their progress after a system failure. This update adds the check-pointing feature, which tracks their progress and allows mdadm to resume these operations after a system failure from the last check point.
BZ#633690
This update adds SAS-SATA feature, which allows the user to disconnect a set of SATA (Serial Advanced Technology Attachment) drives from a SCSI Controller Unit (SCU) and connect it to an Advanced Host Controller Interface (AHCI) and vice versa.
BZ#633692
This update limits the RAID-5 support for volumes on drives attached to the SCU on systems with the X79 chipset (code-named "Patsburg"). It inhibits the creation, assembly, activation, and level migration of RAID 5 volumes on drives attached to the SCU on these systems.
Users are advised to upgrade to this updated mdadm package, which resolves these issues and adds these enhancements.
An updated mdadm package that fixes several bugs is now available for Red Hat Enterprise Linux 6.
The mdadm package contains a utility for creating, managing, and monitoring Linux MD (multiple disk) devices.
Bug Fixes
BZ#723135
When an array was changing level from redundant to non-redundant such as RAID 0, mdadm failed to exit and remained in memory. This caused a variety of issues, including a termination with a segmentation fault when two array transitions were executed sequentially. With this update, a patch has been provided to address this issue, and the monitor is now properly removed from memory after the backward takeover operation.
BZ#723137
If an array size was not properly aligned to the chunk size, the expansion process failed to start and the "New chunk size does not divide component size" error message was returned. This bug has been fixed, and the array alignment is now checked before the expansion process begins.
BZ#723139
Previously, when a volume was created with the "--size" option but without providing a chunk size and a container, mdadm used the "UnSet=65534" option for rounding the volume size before setting the default chunk size. This caused the new volume to be created with an incorrect size. This bug has been fixed, and the volume size is now properly rounded to the chunk size.
BZ#723140
When the expansion process of a RAID 0 volume was restarted, mdadm failed to correctly assemble the array because the critical section could not be restored from a backup file, and the "mdadm: Failed to restore critical section for reshape - sorry" error message was returned. A patch has been provided to address this issue, and now, the RAID 0 expansion process cannot be restarted, thus fixing this bug.
BZ#723141
Previously, mdadm incorrectly calculated reshape start data disks number (0) during the reshape restart operation and used it for calculations. This caused a variety of issues; for example, when two disks were added to the 3-disk RAID 5 array and the array under migration was disassembled and then assembled again, mdadm terminated unexpectedly. A set of patches has been provided to address this issue, and the reshape process is now properly restarted in the described scenario.
BZ#723142
When two arrays were configured in a container and the arrays were reassembled during a rebuild or initialization process, the stored checkpoint for one of the arrays was sometimes lost. Consequently, the restarted process did not use the checkpoint information and started from zero position instead. A patch has been provided to address this issue, and the restarted process now properly continues from the stored checkpoint.
Users of mdadm are advised to upgrade to this updated package, which fixes these bugs.