2.2 Known Issues

This section describes known issues in this update.

Booting UEK R2 as a 32-bit PVHVM Guest

When booting UEK R2 as a 32-bit PVHVM guest, you can safely ignore the kernel message register_vcpu_info failed: err=-38, which might be displayed. (Bug ID 13713774)

Broadcom NetXtreme II 10Gbps Network Adapter Driver

When using the bnx2x driver in a bridge, disable Transparent Packet Aggregation (TPA) by including options bnx2x disable_tpa=1 in /etc/modprobe.conf. (Bug ID 14626070)

Btrfs File System Issues

For a description of the known issues for btrfs with Unbreakable Enterprise Kernel Release 3, see the Oracle Linux Unbreakable Enterprise Kernel Release 3 Release Notes.

For a description of the known issues for btrfs with Unbreakable Enterprise Kernel Release 2 Quarterly Update 5, see the Oracle Linux Unbreakable Enterprise Kernel Release 2 Quarterly Update 5 Release Notes.

Console Appears to Hang when Booting

On some hardware, the console may appear to hang during the boot process after starting udev. However, the system does boot properly and is accessible. A workaround to this problem is to add nomodeset as a kernel boot parameter in /etc/grub.conf. (Bug ID 10094052, 13485328)

Default I/O Scheduler

For the Unbreakable Enterprise Kernel, deadline is the default I/O scheduler.

For the Red Hat Compatible Kernel, cfq is the default I/O scheduler.

Default NFS Mount Options

The default NFS mount option has changed to NFS v4. To mount an NFS v3 volume (the default in Oracle Linux 5), use the following mount options:

-o vers=3,mountproto=tcp

Default Reverse Path Filtering Mode Affects Certain Oracle Products

Oracle Linux 6 defaults to reverse path filtering in strict mode. Some Oracle products and network storage devices work more reliably with reverse path filtering in loose mode. To enable loose mode, issue the following command (where iface is the network interface, for example, eth1).

# sysctl net.ipv4.conf.iface.rp_filter=2

The default setting is 1 for strict mode. (Bug ID 10649976)

DTrace Issues

Enabling FIPS Mode

To make an Oracle Linux Release 6 Update 5 system compliant with Federal Information Processing Standard (FIPS) Publication 140-2, perform the following steps:

  1. Install the dracut-fips package:

    # yum install dracut-fips
  2. Recreate the initramfs file system:

    # dracut -f
  3. Identify either the device file path (device) under /dev of your system's boot device or its UUID (uuid) by using ls -l to examine the entries under /dev/disk/by-uuid.

  4. Add either a boot=device entry or a boot=UUID=uuid entry for the boot device to the kernel command line in /etc/grub.conf.

  5. Add a fips=1 entry to the kernel command line in /etc/grub.conf to specify strict FIPS compliance.

  6. Disable prelinking by setting PRELINKING=no in /etc/sysconfig/prelink.

  7. Remove all existing prelinking from binaries and libraries:

    # prelink -ua
  8. Install the openssh-server-fips and openssh-client-fips packages and their dependent packages:

    # yum install openssh-server-fips openssh-client-fips
  9. Shut down and reboot the system.

Note

If you specify fips=1 on the kernel command line but omit a valid boot= entry, the system crashes because it cannot locate the kernel's .hmac file.

If you do not disable and remove all prelinking, users cannot log in and /usr/sbin/sshd does not start.

(Bug ID 17759117, 17776875)

Error message Following First Reboot on an HP ProLiant Server

You might see a message similar to the following during the first reboot of an HP ProLiant server:

[Firmware Bug]: the BIOS has corrupted hw-PMU resources (MSR 186 is 43003c)

You can safely ignore this message. The functionality and performance of the operating system and the server are not affected.

Incorrect Package Count

Selecting all packages in certain groups during installation might not show the correct package count. (Bug ID 11684244)

iTCO_wdt Errors

If you see the following boot-time dmesg error with the UEK R2:

iTCO_wdt: failed to reset NO_REBOOT flag, device disabled by hardware/BIOS

add the line blacklist iTCO_wdt to /etc/modprobe.d/blacklist-watchdog.

Journal Corruption in Virtualized Systems

On virtualized systems that are built on Xen version 3, including all releases of Oracle VM 2 including 2.2.2 and 2.2.3, disk synchronization requests for ext3 and ext4 file systems result in journal corruption with kernel messages similar to the following being logged:

blkfront: barrier: empty write xvda op failed
blkfront: xvda: barrier or flush: disabled

In addition, journal failures such as the following might be reported:

Aborting journal on device xvda1

The workaround is to add the mount option barrier=0 to all ext3 and ext4 file systems in the guest VM before upgrading to UEK R3. For example, you would change a mount entry such as:

UUID=4e4287b1-87dc-47a8-b69a-075c7579eaf1  /  ext3  defaults  1 1

so that it reads:

UUID=4e4287b1-87dc-47a8-b69a-075c7579eaf1  /  ext3  defaults,barrier=0  1 1

This issue does not apply to Xen 4 based systems, such as Oracle VM 3. (Bug ID 17310816, 17313428)

Kdump Service Configuration

By default, the Kernel Dump service (Kdump) is enabled but not configured. The following boot-time message indicates that Kdump needs to be configured.

kdump: No crashkernel parameter specified for running kernel

To prevent this message from being displayed, use the Kernel Dump Configuration GUI (system-config-kdump) to configure or disable Kdump. (Bug ID 16242031)

When configuring a crashkernel setting for the UEK, only standard settings such as crashkernel=128M@32M are supported. Settings such as crashkernel=auto, which can be used with the Red Hat Compatible Kernel, are not supported by the UEK and cause Kdump to fail to start. (Bug ID 13495212)

Linux Containers 32-bit Support

The Linux Containers package (lxc) is not available for the i386 architecture.

Linux Containers Interoperability with SELinux

The correct operation of containers might require that you completely disable SELinux on the host system. For example, SELinux can interfere with container operation under the following conditions:

To disable SELinux on the host:

  1. Edit the configuration file for SELinux, /etc/selinux/config and set the value of the SELINUX directive to disabled.

  2. Shut down and reboot the host system.

(Bug ID 15967411)

Linux Containers Default Configuration Location

The default location for a container's configuration has changed from /etc/lxc/name to /container/name in lxc 0.8.0 onward.

To start a container that you created with a previous update of Oracle Linux, specify the -f option to lxc-start, for example:

# lxc-start -n ol6u3 -f /etc/lxc/ol6u3/config

To convert an existing container to use the new location:

  1. Move the container's configuration directory to /container/name:

    # mv /etc/lxc/name /container 
  2. Edit the /container/name/config file and change the values of any lxc.rootfs and lxc.mount parameters to refer to /container instead of /etc/lxc.

    For example, if the config file contained the following entries:

    lxc.rootfs = /etc/lxc/example/rootfs
    lxc.mount.entry=/lib /etc/lxc/example/rootfs/lib none ro,bind 0 0
    lxc.mount.entry=/usr/lib /etc/lxc/example/rootfs/usr/lib none ro,bind 0 0
    lxc.mount.entry=/lib64 /etc/lxc/example/rootfs/lib64 none ro,bind 0 0
    lxc.mount.entry=/usr/lib64 /etc/lxc/example/rootfs/usr/lib64 none ro,bind 0 0

    you would change these entries to read:

    lxc.rootfs = /container/example/rootfs
    lxc.mount.entry=/lib /container/example/rootfs/lib none ro,bind 0 0
    lxc.mount.entry=/usr/lib /container/example/rootfs/usr/lib none ro,bind 0 0
    lxc.mount.entry=/lib64 /container/example/rootfs/lib64 none ro,bind 0 0
    lxc.mount.entry=/usr/lib64 /container/example/rootfs/usr/lib64 none ro,bind 0 0

After converting the container, you do not need to specify the -f option to lxc-start. (Bug ID 15967411)

Mellanox ConnectX Drivers

The Mellanox ConnectX core, Ethernet, and InfiniBand drivers are supported only for the x86-64 architecture. (Bug ID 16228063)

mlx4_core Conflicts Between the mlnx_en and ofa Packages

Both the mlnx_en and ofa packages contain mlx4_core. Only one of these packages should be installed. Attempting to install both packages on a single server results in a package conflict error. If you have a Mellanox Ethernet Controller, install mlnx_en. If you have a Mellanox InfiniBand Controller, install ofa. If your system has both controllers, use ofa as it supports both the Ethernet and InfiniBand controllers.

Oracle Clusterware Fails to Start on ASM Storage with SELinux Enabled

If the SELinux policy packages have not been updated recently, Cluster Ready Services (CRS) might fail to start with messages such as the following in /var/log/messages:

SELinux is preventing /usr/lib/oracleasm/oracleasm-instantiate-disk from
associate access on the filesystem DATA1.

The solution is to upgrade the selinux-policy and selinux-policy-targeted packages to ensure that you are running a version no earlier than 3.7.195.0.1.el6_4.5:

# yum update 'selinux-policy*'

After upgrading the packages, reboot the system. (Bug ID 13925445)

Oracle RDBMS Server 11gR2 Preinstall Settings Not Visible After Installation

On an x86-64 system, if you install the pam.i386 package either manually or via a package dependency, and the oracle-rdbms-server-11gR2-preinstall package is also selected, this overwrites the settings for Oracle Database in /etc/security/limits.conf. This is most likely to occur during a Kickstart-automated installation that includes non-standard packages. To restore the settings, run the oracle-rdbms-server-11gR2-preinstall-verify script. (Bug ID 14212822)

Oracle VM 3.0 Guests Crash During Oracle Database Installation

PVHVM guests on Oracle VM 3.0 crash during Oracle Database installation if the value of the maximum memory (maxmem) parameter set for the guest is greater than the amount specified at boot time (memory). To avoid this issue, ensure that the values of the maxmem and memory parameters are the same. This issue has been resolved in Oracle VM 3.1.1. (Bug ID 13396734)

Paravirtualized Drivers in a Hardware Virtualized Guest

The Unbreakable Enterprise Kernel adds support for PV drivers in a HVM guest (PVHVM) on Oracle VM. The default is to present only PV drivers when running in an HVM guest. To run kernel-uek fully hardware virtualized, including the drivers, add the parameter xen_emul_unplug=never to the boot parameters in /etc/grub.conf, for example:

kernel /vmlinuz-2.6.32-300.2.1.el6uek ro root=/dev/VolGroup00/LogVol00 xen_emul_unplug=never

Adding this parameter makes the kernel also present the emulated drivers as previously (for example, the 8139cp network driver).

Post-Installation Anaconda Errors

In certain cases, after successfully completing installation and rebooting the system, it is possible for errors such as the following to occur:

Error in sys.excepthook:
 Traceback (most recent call last):
   File "/usr/lib/python2.6/site-packages/meh/handler.py", line 161, in
 (lambda)
   File "/usr/lib/anaconda/exception.py", line 44, in handleException
   File "/usr/lib/python2.6/site-packages/meh/handler.py", line 106, in
 handleException
   File "/usr/lib/anaconda/gui.py", line 1169, in mainExceptionWindow
 ImportError: No module named ui.gui
14:05:55 CRITICAL: anaconda 11.5.0.47 exception report
Traceback (most recent call first):
  File "/usr/lib64/python2.6/site-packages/gtk-2.0/gtk/_lazyutils.py", line 32,
in __
getattr__
  File "/usr/lib/anaconda/gui.py", line 1453, in keyRelease
    if ((event.keyval == gtk.keysyms.KP_Delete
ImportError: No module named keysyms

These errors can safely be ignored.

RDMA Does Not Load the mlx4_ib Module

If you enable the OFED stack and the RDMA service but the version of the RDMA package is lower than rdma‑3.10‑3.0.2.el6, the RDMA service does not load the mlx4_ib module automatically.

To configure the RDMA service to load the mlx4_ib module at boot time:

  1. Edit /etc/rdma/rdma.conf and set the entry MLX4_LOAD=yes in this file.

  2. To make the change take effect, restart the RDMA service or reboot the system.

Receive Packet Steering Errors

Certain network operations that utilize receive packet steering could cause errors on the system. (Bug ID 11071685)

sched_yield() Settings for the Completely Fair Scheduler

For the Unbreakable Enterprise Kernel, the default setting is kernel.sched_compat_yield=1.

For the Red Hat Compatible Kernel, the default setting is kernel.sched_compat_yield=0.

Setting the Serial Console in a Hardware Virtualized Guest

To set the serial console in a hardware virtualized guest, use following settings in the guest:

Spurious Udev Messages During Failed Path Restoration

If failed paths are restored in a multipath configuration, you might see udevd-work error messages in /var/log/messages. The failed paths are restored despite these messages, which you can ignore. (Bug ID 11682171)

Support for Large Memory 32-bit Systems

Releases of Oracle Linux prior to Oracle Linux 5 supplied a hugemem kernel to allow a system to address up to 64 GB of memory in 32-bit mode. The hugemem kernel is no longer available in Oracle Linux 5 and later releases.

The Unbreakable Enterprise Kernel (UEK) supports a maximum of 16 GB of memory for 32-bit kernels on bare metal and hardware virtualized machine (HVM) systems, and 8 GB for fully paravirtualized machine (PVM) systems. 32-bit PVM guest operating systems must be located in the first 128 GB of physical memory on the host.

The Red Hat Compatible Kernel (RHCK) has the same limitations, except that PVM systems can have up to 16 GB of memory. The limitation of 8 GB for PVM on UEK was chosen for reasons of reliability.

A 32-bit system uses the PAE (physical address extension) memory feature to map physical memory beyond 4 GB into the 32-bit address space that is available to each process. A 64-bit system can address memory beyond 4 GB without requiring an extra layer of memory abstraction.

Oracle Linux on x86-64 includes 32-bit libraries, which allow applications built for both 64-bit and 32-bit Linux to run on the same system. This capability provides scalability to virtually unlimited memory sizes, while retaining the ability to run 32-bit applications. Oracle recommends this configuration for any system with more than 4 GB of memory. (Bug ID 16974301)

udevd Message

A message similar to the following might be recorded in dmesg or /var/log/messages at boot time:

udevd (pid): /proc/pid/oom_adj is deprecated, please use /proc/pid/oom_score_adj instead.

The udevd process uses the deprecated oom_adj kernel interface to prevent it from being killed if the system runs short of memory. You can safely ignore the message as the action still succeeds. To prevent the message from occurring, install the package udev-147-2.42.el6.arch.rpm or higher. (Bug ID 13655071, 13712009)

Unable to Register Oracle Linux Guest with ULN

Registering an Oracle Linux guest running under Virtual Box with the Unbreakable Linux Network (ULN) might fail with a server communication error. The workaround is to run the following command as root on the guest:

# echo "uuid=`uuidgen -t`" >> /etc/sysconfig/rhn/up2date

You can then run uln_register again. (Bug ID 14696776)

Unable to Register with ULN After First Reboot

Following the first reboot after installing Oracle Linux 6, you are prompted to register your system with ULN. If you did not configure your network during the installation, the registration process to ULN cannot succeed. To register your system, log in as root, configure the system's network manually, and run uln_register.

xguest Package Not Installable with SELinux Disabled

If the xguest package fails to install with a PREIN script error, enable SELinux by setting SELINUX=enforcing in /etc/selinux/config, reboot the system, and reinstall the xguest package. (Bug ID 13495388)