1.61. kexec-tools

1.61.1. RHBA-2011:0061: bug fix update

An updated kexec-tools package that fixes various bugs is now available
kexec-tools provides /sbin/kexec binary that facilitates a new kernel to boot using the kernel's kexec feature either on a normal or a panic reboot. This package contains the /sbin/kexec binary and ancillary utilities that together form the userspace component of the kernel's kexec feature.
This update fixes the following bugs:
* The setting of the path line on /etc/kdump.conf was not reflected. Now, this script is corrected. ( BZ#379521)
* Certain systems became unresponsive because Kdump could not handle receipts of dhcp NAK frames. This update resolves this issue. ( BZ#515208)
* Mkdumprd failed because the temporary file was not created when root was read-only. Now, the command "mkdumprd" works as expected. ( BZ#530265)
* Mkdumprd failed when the using a network bridge. Now, kdump works as expected.( BZ#558733)
* Kdump failed to dump vmcore on Nehalem EX machine. Now, kdump dumps vmcore as expected.( BZ#559928)
* Systems could fail to perform a core dump due to DNS problems on AMD64 or Intel64 based hardware on the x86 PAE kernel. This update resolves this issue. ( BZ#559928)
* A reference to deprecated .ssh/authorized_keys2, not .ssh/authorized_keys was hardcoded in kdump. Now, the key is configurable. ( BZ#567708)
* In the kexec initrd environment, the request_module() failed. Now, request_module() loads the requested module. ( BZ#576075)
* Kdump did not save vmcore on machines with 1TB memory. Now, kdump saves vmcore. ( BZ#585979)
* Mkdumprd regenerated the ld.so.cache. This removed any symlinks to shared libs which reside on autofs managed filesystems. Now, mkdumprd restores the ld.so.cache to return the original cache even if network filesystems aren't available. ( BZ#589768)
* Kexec could not dump vmcore through NFS on PowerPC systems due to issues with DNS lookups. Now, dumps to NFS locations succeed. ( BZ#601273)
* The kexec kernel attempted to load inside xen-guest and could panic the central processing unit. Now, kexec-tools prevent the kexec kernel from loading in xen-guest. ( BZ#608322)
* Kdump would not start on ppc servers with 256 Gigbytes of memory enabled. This update fixes this problem. ( BZ#625828)
* The call to kdump_pre was missing from the kdump init script. Now, kdump_pre is part of the kdump init script. ( BZ#640577)
* Kdump saved the vmcore to the Network File System too slow. This update improves the performance. ( BZ#642552)
* The ld.so.cache could not be restored in read-only root. This issue is resolved. ( BZ#642921)
* The ssh key did not propagate. Now, kdump uses '.ssh/authorized_keys' instead of '.ssh/authorized_keys2'. ( BZ#643325)
* Depmod in mkdumprd failed on dom0 with memory bigger than 4 Gigabytes. This update resolves this issue. ( BZ#646791)
* Systems became unresponsive when a forced crash occured due the hugepages=X boot option. Now, the system exits and produces a core. ( BZ#647046)
* The system was not rebooted after dumping. Now, kdump reboots to normal system if no "default" action is available. ( BZ#647627)
* Due to DNS problems nfs dumping failed on dom0. This update resolves this problem. ( BZ#648416)
* Errors occured during ssh dumping. Now, ssh dumping works as expected. ( BZ#648766)
* While loading shared libraries: libc.so.6, ssh dumping failed because ifup did not have all the variables for lo/inet. This update resolves this issue. ( BZ#648771)
* Warning messages echoed when starting kdump service. Now, kdump service behaves as expected. ( BZ#652500) * The system would reboot when dhcp failed to read IPs. Now, the system enters use-space and saves the vmcore. ( BZ#654997)
* The vmcore was wrongly named vmcore.flat due problems with 'core_collector cp'. This update resolves this issue. ( BZ#657198)
This update also adds the following enhancement:
* This update adds the opption "poweroff" to 'default' in /etc/kdump.conf. ( BZ#543336)
All users are advised to upgrade to this updated kexec-tools package, which resolves these issues and adds this enhancement.

1.61.2. RHBA-2010:0776: bug fix update

An updated kexec-tools package that fixes a bug is now available.
The kexec-tools package provides /sbin/kexec binary that facilitates a new kernel to boot using the kernel's kexec feature either on a normal or a panic reboot.
This updated kexec-tools package fixes the following bug:
* Previously, "/sbin/kexec" could run out of buffer on a PowerPC, when the PowerPC's memory was large, due to the buffer size limit for devtree. This caused kdump to fail to start. With this update, the buffer size is increased, thus, segmentation faults no longer occur and kdump is able to start as expected. ( BZ#639303)
All users of kexec-tools are advised to upgrade to this updated package, which resolves this issue.

1.61.3. RHBA-2010:0438: bug fix update

An updated kexec-tools package that fixes a bug that prevented kexec from capturing a vmcore on 64-bit systems with more than 1TB of RAM is now available.
kexec-tools provides the /sbin/kexec binary that facilitates a new kernel to boot using the kernel's kexec feature either on a normal or a panic reboot. This package contains the /sbin/kexec binary and ancillary utilities that together form the userspace component of the kernel's kexec feature.
This update addresses the following issue:
* the physical address space supported by the Red Hat Enterprise Linux 5 kernel was out of sync with the physical address space the kexec binary assumed the kernel supported. As a consequence, kexec could not successfully capture a vmcore on AMD64 or Intel 64 systems with more than 1TB of RAM. This update brings the kernel and kexec utility view of the supported memory boundary into sync, obviating the problem. ( BZ#590547)
All users are advised to upgrade to this updated kexec-tools package, which fixes this bug.

1.61.4. RHBA-2010:0375: bug fix update

An updated kexec-tools package that fixes various bugs is now available.
kexec-tools provides /sbin/kexec binary that facilitates a new kernel to boot using the kernel's kexec feature either on a normal or a panic reboot. This package contains the /sbin/kexec binary and ancillary utilities that together form the user-space component of the kernel's kexec feature.
This updated kexec-tools package fixes the following bug:
* systems running a 32-bit kernel with the PAE (Physical Address Extension) feature enabled restrict the range of accessible memory to 64 GB, even when the system contains more physical RAM than 64 GB. On such a system, the kernel exports all of the detected memory via /proc/iomem, not just the amount of accessible memory. The kexec user-space tools use /proc/iomem to determine the amount of system memory in order to format its vmcore file so that all memory is accessible. However, on systems with more than 64 GB of RAM, the value read by kexec from /proc/iomem did not represent the amount of actual memory used by the system, which resulted in /proc/vmcore read errors when the kdump mechanism activated.
This update restricts the size of the vmcore file on 32-bit systems to 64 GB, thus preventing /proc/vmcore read errors due to an incorrectly-sized vmcore file. ( BZ#582349)
All users of kexec-tools are advised to upgrade to this updated package, which resolves this issue.