1.22. device-mapper-multipath
1.22.1. RHEA-2011:0074: bug fix and enhancement update
Updated device-mapper-multipath packages that fix several bugs and add various enhancements are now available.
The device-mapper-multipath packages provide tools to manage multipath devices using the device-mapper multipath kernel module.
Bug fixes:
* The kernel driver of device-mapper multipath appeared to be old and SAN boot could fail, because the version testing of the driver was incorrect. Version testing is now implemented correctly. ( BZ#517951)
* On a non-disruptive upgrade (NDU), all paths of EMC Symmetric arrays failed and the application suddenly terminated. The default configurations for EMC Symmetric are added and the issue no longer occurs. ( BZ#562967)
* device-mapper multipath failed to update its maps after a path state transition. multipathd now automatically updates the path groups when a path priority changes. ( BZ#566685)
* Because of a udev race, partitions on multipath devices were not always created. The udev rules now wait for device-mapper multipath to finish the device creation. ( BZ#568127)
* If a multipath device configured with queue_if_no_path with no working paths was created while booting, the machine hung. multipath no longer enables queuing before multipathd is started. ( BZ#579789)
* If a default configuration used /bin/true for the priority callout, booting from a multipathed root file system stopped responding. Now the default values of the respective configurations are set to "none" and booting succeeds. ( BZ#580729)
* To resize a multipath device, you had to run "multipathd -k'resize map [mapname]'", as running "multipath" caused it to get stuck in the SUSPENDED state. In this state, all I/O to the device hung. You can now resize the device with either command. ( BZ#584742)
* The "show config" command displayed only values differing from the default values. Now, the command prints the complete configuration. ( BZ#585225)
* If you added and removed a path in quick succession, the path was removed while multipathd was still using it, resulting in a segmentation fault. multipathd no longer crashes. ( BZ#597789)
* If all paths to a device failed and were recovered, in some cases multipath was unable to recover the path immediately. multipath now detects recovered paths. ( BZ#599053)
* Since the HWTABLE cannot be overridden, the "defaults" section of the multipath.conf man page no longer implies that anything in the section becomes default and overrides the implied settings. ( BZ#599686)
* The kernel could send a spurious remove uevent for a multipath device and a valid multipath device was removed. uevents are now sent only after a device is removed. ( BZ#614966)
* Restoring paths on a group_by_prio multipath device could cause the pathgroups to be misconfigured. Multipathd now refreshes all path priorities after paths are restored, which solves the issue. ( BZ#660238)
Enhancements:
* CCISS devices are now supported. ( BZ#484419)
* Multipath for storage devices mounted as read-only with the 'nodev' option is added. ( BZ#559852)
* The multipath and multipathd commands print warning messages in case of an invalid configuration parameter in multipath.conf. ( BZ#560892)
* The default configuration for IBM Virtual SCSI ALUA, HP EVA 3000/5000 and P2000, EUROLOGC FC2502 devices is provided. ( BZ#647358, BZ#565579,
* The "pg_prio_calc" option was added to multipath.conf default options. By default, the option is set to "sum" and group priority is calculated as the sum of its path priorities. If set to "avg", multipath calculates priorities using the average priority of the paths in the group. ( BZ#570513)
* The "log_checker_err" option was added to the multipath.conf defaults option. By default, the option is set to "always" and a path checker error is logged continuously. If set to "once", multipathd logs the path checker error once at logging level 2. Any later errors are logged at level 3, until the device is restored. ( BZ#574813)
* Virtio block devices can be multipathed. ( BZ#576600)
Users should upgrade to these updated packages, which resolve these issues and add these enhancements.
1.22.2. RHBA-2010:0728: bug fix update
Updated device-mapper-multipath packages that fix a bug are now available for Red Hat Enterprise Linux 5.
The device-mapper-multipath packages provide tools to manage multipath devices by giving the "dm-multipath" kernel module instructions on what to do, as well as by managing the creation and removal of partitions for Device-Mapper devices.
These updated packages provide a fix for the following bug:
* Due to a race condition between udev and Device-Mapper, kpartx may have attempted to create partitions on a multipath device before it was completely initialized, causing no Device-Mapper partitions to be created. With this update, udev now waits for the multipath device to be completely initialized before attempting to create partitions on it. ( BZ#634903)
Users are advised to upgrade to these updated device-mapper-multipath packages, which resolve this issue.
1.22.3. RHBA-2010:0696: bug fix update
Updated device-mapper-multipath packages that fix a bug are now available for Red Hat Enterprise Linux 5.
The device-mapper-multipath packages provide tools to manage multipath devices by giving the "dm-multipath" kernel module instructions on what to do, as well as by managing the creation and removal of partitions for Device-Mapper devices.
These updated packages provide a fix for the following bug:
* when multipathd occassionally received spurious remove uevents from the kernel, multipathd incorrectly removed the multipath device. Since remove uevents are only supposed to be generated after a device has been already removed, multipath no longer removes a multipath device when it receives a remove uevent. Instead, it simply cleans up its internal state if the device has been removed. ( BZ#624772)
Users are advised to upgrade to these updated device-mapper-multipath packages, which resolve this issue.
1.22.4. RHBA-2010:0532: bug fix update
Updated device-mapper-multipath packages that fix multiple bugs are now available.
The device-mapper-multipath packages provide tools to manage multipath devices by giving the "dm-multipath" kernel module instructions on what to do, as well as by managing the creation and removal of partitions for Device-Mapper devices.
This update fixes the following bugs:
* due to the inability of multipathd (a path monitoring daemon) to distinguish between deleted and recently added but not yet fully initialized devices, it used to take more than a minute to remove a dead path and reload the device with a new one. This has been fixed and multipathd now removes the deleted devices immediately. ( BZ#603616)
* previously, to resize a multipath device, users had to run the command "multipathd -k'resize map <mapname>'", as running "multipath" instead caused the multipath to become stuck in the SUSPENDED state. In this state, all input/output to the device hung. Now users can resize the device with either command, and the multipath device will no longer get stuck in the SUSPENDED state. ( BZ#607487)
* adding and removing a path in quick succession could have caused it to be removed while the multipathd was still using it, which used to lead to a segmentation fault. This has been resolved, and multipathd no longer crashes when the path is added and quickly removed again. ( BZ#607911)
Users are advised to upgrade to these updated device-mapper-multipath packages, which resolve these issues.
1.22.5. RHBA-2010:0415: bug fix update
Updated device-mapper-multipath packages that fix two bugs are now available.
The device-mapper-multipath packages provide tools to manage multipath devices by giving the device-mapper multipath kernel module instructions on what to do, as well as by managing the creation and removal of partitions for device-mapper devices.
This update addresses the following bug:
* During bootup, multipath was setting up queue_if_no_path on devices that asked for it before multipathd was started. If paths to the device existed, but were all unusable, multipath would queue all IO until multipathd started up. This caused kpartx to hang when trying to read the partition table off the device. The end result was that the entire machine would hang during bootup. Multipath now does not enable queueing on newly created devices if multipathd has not started up yet. This can be overridden with the new -q option. Multipathd will enable queuing on the appropriate devices when it starts up. ( BZ#584344)
Users are advised to upgrade to these updated device-mapper-multipath packages, which resolve this issue.