- Previously, anaconda did not recognize a DVD ISO image written to a USB drive as a source repository for installation because this device does not have partitions. Consequently, the ISO acted like a
boot.isoand it was not possible to install packages included in it. With this update, anaconda has been modified to include devices with ISO 9660 formatting, and to configure any device as a source repository if this device contains the
/repodata/repomd.xmlfile. As a result, anaconda now recognizes ISO on USB as expected.
- Prior to this update, the anaconda
loadercommand created the
/etc/sysconfig/networkfile by renaming a new temporary file, which did not trigger the NetworkManager's inotify mechanism. Consequently, a hostname set by the
network --hostnamekickstart option could be overridden by NetworkManager with hostname obtained though DHCP or DNS. With this update,
loaderhas been modified to write new values directly into
/etc/sysconfig/network. As a result, NetworkManager now accepts the hostname value specified in this file.
- Previously, anaconda did not attempt to use another loop device if the firstly selected one was already in use. Consequently, HDD ISO installation failed if loop devices were used in the kickstart %pre section. With this update, anaconda has been modified to use another loop device if the first one is already in use. As a result, HDD ISO installation works as expected.
- With this update, the
list-harddrivescommand has been modified not to list the
/dev/srXdevices in its output.
- With this update, several typographical errors have been corrected in the About LVM dialog.
- When a different set of disks was used for the
part --ondiskcommands, a backtrace was returned. Consequently, installation did not finish successfully. With this update, only one set of disks is used with these commands. User must specify multiple disks with a single
clearpartcommand, otherwise only the last
clearpart --drivesarguments is used.
- Prior to this update, when partitioning was incorrectly specified, the
No free space errormessage was incorrectly shown instead of the appropriate
No free slotsdialog. With this update, the correct error message is displayed in case of incorrectly specified partitioning.
- Previously, anaconda in rescue mode unmounted the source ISO before searching for the
.discinfofile. Consequently, the
stage2parameter was loaded twice, increasing the boot time. With this update, anaconda has been modified to skip the check for
.discinfoin rescue mode. As a result,
stage2is only loaded once, as expected.
- Previously, when re-installing the system with already configured LVM raid1 volumes, anaconda terminated unexpectedly. This bug has been fixed, and anaconda no longer crashes in the aforementioned scenario.
- When a kickstart upgrade was performed on IBM System z architectures, anaconda shut down the system instead of rebooting even though the
rebootcommand was present in the kickstart configuration. Consequently, a manual reboot was required. This update adds support for kickstart upgrades on System z, thus fixing this bug.
- Previously, when installing Red Hat Enterprise Linux 6 on a system with multiple disks and one or more of these disks contained the PPC PReP Boot partition, anaconda created an empty PPC PReP Boot partition on the selected installation disk, but stored necessary boot files in the already existing PPC PReP Boot. Consequently, the system failed to boot after the installation. With this update, anaconda has been modified to use the correct PPC PReP Boot for boot files, thus fixing this bug.
- The algorithm for calculating the swap size did not take into account the amount of space used for the installation. Consequently, even on small disks the installer created big swap space often leaving only insufficient amount of space for the rest of the system. This algorithm has been modified to register the amount of space used for the installation. As a result, smaller (10% of used disks' space) swap is created on machines with small disks leaving more space for the rest of the system.
- Previously, anaconda did not create partitions larger than 16TB on
XFSfilesystems. This bug has been fixed, and the official limit of 100TB is now used as accepted.
- Prior to this update, the
autopartcommand did not function correctly with already defined prepboot partitions. Consequently, when using a
kickstartfile that contained the
partcommand defining a prepboot partition followed by
autopart, anaconda terminated unexpectedly with a segmentation fault. With this update,
autoparthas been modified to work correctly in the aforementioned configuration. As a result, the installation continues as expected.
- Previously, when configuring network devices within the anaconda GUI, devices using the FCoE network technology were automatically set not to be controlled by the NetworkManager. Consequently, NetworkManager disabled these devices, causing previously connected FCoE SAN disks to disappear from the GUI. This bug has been fixed, and editing network device configuration in the GUI no longer disconnects previously set FCoE devices.
- Previously, anaconda did not return a warning message when using a raw partition for the / mount point without creating a new file system. With this update, anaconda has been modified to display a warning message in such scenario.
- Under certain circumstances, when managing partitions with the anaconda GUI, an unexpected loss of window focus occurred. With this update, the parent window setting has been modified, thus fixing this bug.
- With this update, several typographic and translation errors have been corrected in the Japanese locale in anaconda.
- Previously, anaconda created the
/etc/zipl.confconfiguration file using a set of default kernel parameters regardless of whether a fresh install or upgrade was performed. Consequently, kernel parameters added to
/etc/zipl.confby users were lost when upgrading IBM System z systems with anaconda. This update adds support for boot loader upgrades for systems with System z architecture. As a result, kernel parameters added by users to
/etc/zipl.confare preserved in the aforementioned scenario.
- Prior to this update, when attempting to install conflicting packages with the anaconda GUI, a misleading warning message was displayed. With this update, this message has been modified to inform about the package conflict.
- Physical-extents size less than 32MB on top of an MD physical volume leads anaconda to problems with calculating the capacity of a volume group. To work around this problem, use a physical-extent size of at least 32MB or leave free space (with size equal to doubled size of the physical-extent) when allocating logical volumes.
- Before proceeding to the package installation phase, anaconda did not check if the core package group was available in selected repositories. If this group was not present, the installation terminated unexpectedly. With this update, anaconda has been modified to check for the presence of core. As a result, a warning message is displayed when core is not available, and installation no longer crashes.
- Previously, when configuring a VLAN network device, such as eth0.171, during the installation, the same configuration was incorrectly applied also for its parent device. Consequently, the VLAN parent device, such as eth0, was incorrectly configured during the installation. The bug has been fixed, and the VLAN device configuration is now applied correctly.
- Under certain circumstances, kernel command-line entries created by anaconda and passed to GRUB did not work correctly. Consequently, in multi-path configuration, the Boot File System (BFS) terminated unexpectedly when the last FCoE interface specified in kernel command was not on-line. With this update, the form of kernel command-line entries has been modified, and BFS no longer fails in the aforementioned scenario.
- Previously, the
/etc/multipath/bindingsfile had incorrect SElinux context after installation. This bug has been fixed, and
/etc/multipath/bindingsis now installed with correct SElinux context.
- Prior to this update, the generated
kickstartfile did not contain correct network commands for VLAN interfaces. Consequently, these commands were not reusable during the installation. This bug has been fixed, and the generated
kickstartnow contains reusable network commands.
- By default, the AMD IOMMU driver is disabled in Red Hat Enterprise Linux 6 for stability reasons. However, when IOMMU is expected to be present for trusted boot, this driver is needed. With this update, anaconda has been modified to enable AMD IOMMU in the kernel boot parameters when the
tbootpackage is installed. MD IOMMU is enabled when trusted boot is in use and AMD IOMMU specifications are present and enabled in the BIOS. To revert these settings, users may remove the "amd_iommu=on" kernel parameter if stability issues are encountered.
- Previously, it was not possible to blacklist the usb-storage module during the installation of Red Hat Enterprise Linux 6. This bug has been fixed, and usb-storage can now be blacklisted without complications.
- Under certain rare circumstances, the dasd_eckd_mod driver was not loaded during linuxrc.s390 installation and anaconda became unresponsive. With this update, a patch has been applied to prevent this problem.
- Previously, bond network devices were activated only in the early stage of installation. Consequently, bond devices configured by network commands in the
stage2file were not activated. This behavior has been changed and bond devices can now be activated also in later stages of installation.
- Previously, anaconda loaded certain required packages multiple times during installation. Consequently, the dependency solving took a long time, growing with number of disks and file systems. With this update, anaconda has been modified to use a more efficient way of selecting packages, thus reducing the time spent on dependency solving.
- With this update, anaconda no longer requires the fcoe-utils package for installation on the IBM System z architectures.
- Prior to this update, anaconda limited swap size to 10 % of disk space even if
--hibernationoption was used in the
kickstartfile. With this update, anaconda has been modified to accept the
--hibernationoption, and swap size is no longer limited to 10% of disk space when this option is specified.
- Due to an incorrect setting in the
/etc/ssh/sshd_config.anacondaconfiguration file, the
sshddaemon did not start during installation on IBM System z architectures in FIPS mode. Consequently, the installation was not successful. This bug has been fixed, and
sshdnow runs as expected during installation in FIPS mode.
- Prior to this update, multipath devices were not listed during installation in VNC mode. This bug has been fixed, and these devices are now listed properly.
- Devices directly formatted with a file system without any partitions are not supported in Red Hat Enterprise Linux 6. Previously, anaconda did not verify if devices meet this condition. Consequently, when attempting to create a new partition on such unsupported device, anaconda terminated unexpectedly. With this update, anaconda has been modified to check if the device is unpartitioned and to abort partitioning in such case, thus preventing the crash.
- Previously, a bug in the zipl boot loader caused a runtime error in anaconda. Consequently, the IBM System z architectures with rootfs on iSCSI LUN failed to boot after ananaconda upgrade from Red Hat Enterprise Linux 6.4 to 6.5. This bug has been fixed, and the failed booting no longer occurs after system upgrade.
- Due to an outdated FCoE detection for Broadcom adapters in anaconda, the system was unable to boot after OS after FCoE BFS installation on HP systems. With this update, anaconda has been modified to correctly detect FCoE on Broadcom adapters, and the boot problems no longer occur in the aforementioned scenario.
- Under certain circumstances, after an upgrade from Red Hat Enterprise Linux 6.4 to 6.5, the IBM System z system did not boot from the correct storage device. This bug has been fixed, and System z systems now boot from the correct device after upgrade.
- Certain adapters, such as 10GBaseT Twin Pond require longer time to link up. This time often exceeded the timeout limit of the fipvlan tool used by the installer. Consequently, adding FCoE targets in the GUI failed by timing out. With this update, the timeout limit of fipvlan has been raised. As a result, FCoE target is now added successfully regardless of adapter type. Nevertheless, to view the added device in the GUI, user has to go two screens back to the language selection and then forth.
- Previously, the list of FCoE LUNs disappeared from the SAN Devices tab in anaconda after adding a second adapter during installation of the Specialized Storage BFS. This bug has been fixed, and the list is now displayed correctly during the installation.
- Prior to this update, anaconda incorrectly extracted partition names for NVMe devices. Consequently, the boot loader installation failed on NVMe devices. This bug has been fixed, and NVMe devices are now installed successfully.
- This update adds more flexible support for disk references within the
--driveorderoption in the kickstart boot loader. It is now possible to specify disks that use the
/dev/disk/by-*/folders as arguments for
- This update adds the
--ipv6gatewayoption to the kickstart network command, which allows to specify a default IPv6 gateway. Now, both IPv4 and IPv6 default gateways can be specified in network kickstart command using
- With this update, a partition size check has been added to anaconda to ensure that the boot partition on x86 architectures is always less than 2TB, which is required by the GRUB boot loader.
- With this update, anaconda has been modified to allow the DDNS method in the installer. If a
hostnameis specified in the kickstart configuration of a network device that uses the DHCP protocol, this
hostnameis passed to the dhclient utility.