1.26. device-mapper-multipath

Updated device-mapper-multipath packages that fix several bugs and add various enhancements are now available for Red Hat Enterprise Linux 5.

The device-mapper-multipath packages provide tools to manage multipath devices using the device-mapper multipath kernel module.

This update fixes the following bugs:

* Prior to this update, when viewing partition information from a file, kpartx used a loop device which was not released by kpartx afterwards. With this update, this erroneous behavior has been fixed so that kpartx now correctly releases a loop device as expected. (BZ#578109)

* After a thread of the multipathd service acquired its resource locks, the thread did not check if it had been canceled before accessing the resource. This caused multipathd to occasionally terminate unexpectedly with a segmentation fault on shutdown if the thread attempted to access the resource after the thread was canceled. With this update, the bug has been fixed so that the multipathd thread now checks if it is canceled before accessing its resource. (BZ#639429)

* When the multipathd service started, it created a separate namespace for itself, and unmounted all unnecessary disk-backed file systems. Since /ram file systems are not considered to be disk-backed, they should have not been unmounted by multipathd, however multipathd unmounted them. This update corrects this undesired behavior so that the /ram file systems are no longer unmounted. (BZ#663179)

* Prior to this update, the multipathd service did not check if a value was entered for an option in the /etc/multipath.conf configuration file before attempting to read the value. As a result, multipathd terminated unexpectedly when an option without any value was found in /etc/multipath.conf. With this update, the bug has been fixed so that multipathd no longer crashes. (BZ#675369)

* If the last path of a device was deleted while the multipathd service was trying to reload the device map, or if a ghost path failed, multipathd did not always switch into recovery mode. As a result, multipath devices were not able to recover I/O operations in setups that were supposed to temporarily queue I/O if all paths were unavailable. This update resolves both of these problems; multipath now correctly recovers I/O operations as configured. (BZ#677821)

* Prior to this update, there was a spelling mistake found in the "invalid keyword" error message. The spelling mistake has been fixed in this update. (BZ#676165)

As well, this update adds the following enhancements:

* This update introduces two new defaults options in the /etc/multipath.conf configuration file: "fast_io_fail_tmo" and "dev_loss_tmo". The "fast_io_fail_tmo" option controls how long the SCSI layer waits after a SCSI device fails before failing back the I/O. This option can be set to "off" or any number less than the "dev_loss_tmo" option. The "dev_loss_tmo" option controls how long the SCSI layer waits after a SCSI device fails before marking it as failed. The default values for these options are set by the SCSI device drivers. (BZ#672575)

* This update introduces a new defaults section parameter for the /etc/multipath.conf configuration file: the "file_timeout" parameter. This parameter controls how many seconds the multipathd service will wait for a necessary file to appear while setting up a multipath device. The default value is 90 seconds. (BZ#627911)

* This update introduces the default configuration for multiple new HP storage array products. (BZ#502813)

All users of device-mapper-multipath should upgrade to these updated packages, which fix these bugs and add these enhancements.