This document covers the following topics regarding Oracle Linux 6.
For the latest updates please refer to the online version of release notes available at:
http://oss.oracle.com/ol6/docs/
NOTE: An upgrade of Oracle Linux from a beta release is not supported. Further, an in-place upgrade between major versions of Oracle Linux is not supported. Oracle does not recommend an upgrade from earlier major versions of Oracle Linux even though anaconda provides an option to do this upgrade.A fresh installation is strongly recommended rather than a system upgrade between major versions. Customers who want to use new features in Unbreakable Enterprise Kernel can do so without upgrading to Oracle Linux 6 as Oracle Linux 5.6 already includes Unbreakable Enterprise Kernel.
Oracle Linux 6 includes many new features, including:
The ext4 filesystem is installed by default.
Oracle Linux 6 includes XFS as an optional filesystem.
Ftrace is a tracing framework for analyzing performance and latency in the kernel.
The performance counter subsystem keeps track of hardware and software events without affecting performance and enables you to do tracing and performance analysis. Included is a tool called perf for analysis.
Powertop is a new user space tool that helps you reduce server power usage by identifying power hungry processes and by providing recommendations to lower power consumption.
Latency can be defined as follows: an applications wants to run and execute code, but is blocked by the kernel because some required resource is not available. LatencyTOP is a Linux tool for software developers (both kernel and userspace), aimed at identifying where system latency occurs, and what kind of operation or action is causing the latency to happen. By identifying this, developers can then change the code to avoid the worst latency hiccups.
Oracle Linux 6 not longer contains up2date for access to Unbreakable Linux Network. Instead packages are managed using Yum. To register with ULN, use the following command:# uln_register
The following packages are modified from the upstream release. All changes are trademark and look/feel related unless otherwise noted below under the specific package.
The following packages have been removed.
The following packages have been added to the base release
Note:By default, both the Unbreakable Enterprise Kernel and the Red Hat Compatible Kernel are installed on the x86_64 (64 bit) platform, and the system boots the Unbreakable Enterprise Kernel by default. If needed, /etc/grub.conf can be modifed to make the system boot with the Red Hat Compatible Kernel by default.
The Unbreakable Enterprise Kernel is based on the upstream kernel 2.6.32.21 stable source tree with additional performance improvements, including:
The Unbreakable Enterprise Kernel is available for the x86_64 architecture and is the default kernel after installation.
OFED implements Remote Direct Memory Access (RDMA) and kernel bypass mechanisms to deliver high-efficiency computing, wire-speed messaging, ultra-low microsecond latencies and fast I/O for servers, block storage and file systems.
The Unbreakable Enterprise Kernel includes the OCFS2 1.6 kernel module. New features include:For more details, see the OCFS2 1.6 User's Guide
- JBD2 support
- Extended attributes
- POSIX ACLs
- Security attributes
- Metadata checksums
- Indexed directories
- REFLINKs
The Linux data integrity framework (DIF) enables applications or kernel subsystems to attach metadata to I/O operations, allowing devices that support DIF to verify the integrity before passing them further down the stack and physically committing them to disk. Data Integrity Extensions or DIX is a hardware feature that enables exchange of protection metadata between host operating system and HBA.
The Unbreakable Enterprise Kernel is tickless. In the tickless kernel, timer interrupts are performed on demand rather than at a predetermined frequency. This allows CPUs to stay in a low power state when the system is idle, reducing overall power consumption
TCG can track and group processes into user-defined cgroups so that the operating system can treat them as whole and perform scheduling, accounting, and resourceallocation accordingly. For example, using TCG, you can associate a set of CPU cores and memory nodes to a group of processes that makeup an application or a group of applications. This enables subsetting larger systems, more fine grained control over CPU and memory, and isolation of applications.
The performance counter subsystem keeps track of hardware and software events without affecting performance and enables you to do tracing and performance analysis. Included is a tool called perf for analysis.
The kernel block layer will detect devices that claim to be fast,solid state storage and tune itself accordingly. The result of this detection can be found in:/sys/block/xxx/queue/rotationalWhere xxx is the block device. Echoing a 0 or a 1 into this file will force the value to off or on. When assuming a device is an SSD, the block layer will try harder to immediately dispatch the IO to the device.
IO affinity ensures processing of a completed IO is handled by the same CPU that initiated the IO. It can have a fairly large impact on performance, especially on large NUMA machines. IO affinity is turned on by default, but it can be controlled via the tunable in /sys/block/xxx/queue/rq_affinity. For example, the following will turn IO affinity on:echo 1> /sys/block/sda/queue/rq_affinity
RPS distributes the load of received packet processing across multiple CPUs. This solution allows protocol processing (e.g. IP and TCP) to be performed on packets in parallel. This removes a bottleneck when a single core is saturated processing network interrupts. To enable receive packet steering for the interface xxx, place a CPU mask into/sys/class/net/xxx/queues/rx-0/rps_cpusThe cpu mask takes the same form as the masks for the taskset command. For example:echo 0x55> /sys/class/net/eth0/queues/rx-0/rps_cpus
fallocate() is a new system call which will allow applications to preallocate space to any file(s) in a file system. Applications can get a guarantee of space for particular file(s) - even if later the system becomes full. Using this method of allocation can dramatically speed up the creation of large files such as those used for virtual machine images.
Selecting all packages in certain groups during installation might not show the correct count.
On certain systems, kdump will not work properly. This has been seen on P400i controllers using cciss and hpsa drivers.
An application using hugepages may see a warning message like "Using mlock ulimits for SHM_HUGETLB deprecated." To avoid this warning, the application should be configured to use CAP_IPC_LOCK or the process (e.g. Oracle) should be added to the hugetlb_shm_group.
Oracle Linux 6 defaults to strict reverse path filtering. Some Oracle products and network storage devices work more reliably when using loose reverse path filtering. To enable loose mode reverse path filtering, issue the following command (assuming you are changing settings for the network interface called eth1). The default setting is 1.sysctl net.ipv4.conf.eth1.rp_filter=2
Certain network operations that utilize receive packet steering could cause errors on the system.
The sample multipath.conf included in Oracle Linux 6 has an incorrect entry. The path_checker can't be set in the multipath section. Instead, it must be set in the default or device sections.
When failed paths are restored in a multipath configuration, you may see udevd-work error messages in /var/log/messages. The failed paths do get restored despite these messages.
The default NFS mount option has changed to NFS v4. To mount an NFS volume using NFS v3 (the default in Oracle Linux 5), use the following mount options:-o vers=3,mountproto=tcp
To set the serial console a hardware virtualized (HVM) guest use following settings in the guest:
- On the kernel boot line in grub.conf, add:
console=tty0 console=ttyS0,57600n8- add the following to /etc/securetty
ttyS0- add the following to /etc/sysconfig/init
ACTIVE_CONSOLES=/dev/tty[0-6]
As a workaround, install the kernel in an HVM guest without paravirtualized drivers.
After an install of Oracle Linux 6, the screens after the first reboot will guide you to register your system to the Unbreakable Linux Network (ULN). If you did not configure your network during the initial installation, the registration process to ULN will not succeed. In this case, you should configure your network by running NetworkManager first (as root). Then restart the ULN registration by running uln_register (as root).
On some hardware, the console may appear to hang during the boot process after starting udev. But the system does boot up properly and is accessible. A workaround to this problem is to add nomodeset as a kernel boot parameter in /etc/grub.conf:
kernel /vmlinuz-2.6.32-100.28.5.el5.x86_64 ro root=/dev/sd0 nomodeset
As a workaround, when creating a new VM in VSphere, do not pick Red Hat Enterprise Linux 6 x86-64 as the OS type but use Oracle Linux 5 x86-64 (or Red Hat Enterprise Linux 5). ESX will then expose the LSI Logic SCSI controller in the VM and the 2.6.32-100.28.* kernel will see the devices properly.
For the Unbreakable Enterprise Kernel, the default IO scheduler is the 'deadline' scheduler.
For the Red Hat Compatible Kernel, the default IO scheduler is the 'cfq' scheduler.
For the Unbreakable Enterprise Kernel, kernel.sched_compat_yield=1 by default.
For the Red Hat Compatible Kernel, kernel.sched_compat_yield=0 by default.
In some instances, large IO loads on NFS that use mmap may fail and cause the kernel to panic with
a message like:
kernel BUG at mm/truncate.c:469!
When kdump starts, you may see the following warning message:
WARNING: No modules scsi_mod found for kernel 2.6.32-100.28.5.el6.x86_64, continuing anyway
This warning can be ignored. The kdump service does get started and a vmcore does get generated.
Configuring the system with an encrypted disk is currently not supported with the Unbreakable Enterprise Kernel.
The following Technology Preview features are currently not supported under Oracle Linux 6 and may not be functionally complete:
These features are not suitable for production use. However, these features are included to provide the feature with wider exposure.