mirror of
https://github.com/clearlinux/clear-linux-documentation.git
synced 2026-11-22 03:22:45 +00:00
Merge branch 'master' of github.com:clearlinux/clear-linux-documentation into rtd-theme
Signed-off-by: Kevin Putnam <kevin.putnam@intel.com>
This commit is contained in:
+2
-2
@@ -1,7 +1,7 @@
|
||||
.. _faq:
|
||||
|
||||
Frequently Asked Questions (FAQ)
|
||||
################################
|
||||
FAQ
|
||||
###
|
||||
|
||||
Below is a list of commonly asked questions with answers sourced from the
|
||||
|CL-ATTR| team and `Clear Linux community forums`_.
|
||||
|
||||
@@ -70,8 +70,8 @@ Confirm network connection
|
||||
==========================
|
||||
|
||||
Confirm there is a network connection before launching the installer.
|
||||
Choose a method: `Wired Connection`, or `WiFi Connected`.
|
||||
This guide shows an example of a **Wired Connection**.
|
||||
Choose a method: `Wired Connection`, or `WiFi Connected`. This guide shows
|
||||
an example of a **Wired Connection**.
|
||||
|
||||
#. In the upper right of the top menu bar, select the square icon to view
|
||||
Network settings, shown in Figure 2.
|
||||
@@ -243,6 +243,16 @@ Use this method to safely install |CL| on media with available space, or
|
||||
alongside existing partitions, and accept the `Default partition schema`_.
|
||||
If enough free space exists, safe installation is allowed.
|
||||
|
||||
.. note::
|
||||
|
||||
|CL| allows installation alongside another OS. Typically, when you boot
|
||||
your system, you can press an `F key` to view and select a bootable
|
||||
device or partition during the BIOS POST stage. Some BIOSes present the
|
||||
|CL| partition, and you can select and boot it. However, other
|
||||
BIOSes may only show the primary partition, in which case you will not be
|
||||
able boot |CL|. Be aware of this possible limitation.
|
||||
|
||||
|
||||
Destructive Installation
|
||||
------------------------
|
||||
|
||||
|
||||
@@ -3,7 +3,7 @@
|
||||
Run |CL-ATTR| as a KVM guest OS
|
||||
###############################
|
||||
|
||||
This section explains how to run |CL-ATTR| in a virtualized environment using
|
||||
This guide explains how to run |CL-ATTR| in a virtualized environment using
|
||||
:abbr:`KVM (Kernel-based Virtual Machine)`.
|
||||
|
||||
Install QEMU-KVM
|
||||
@@ -13,12 +13,7 @@ Install QEMU-KVM
|
||||
`Intel®Virtualization Technology for Directed I/O`_ (Intel® VT-d) in the
|
||||
host machine’s BIOS.
|
||||
|
||||
#. Log in, open a terminal emulator, and get root privilege on the host
|
||||
machine:
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
sudo -s
|
||||
#. Log in and open a terminal emulator.
|
||||
|
||||
#. Install `QEMU*-KVM` on the host machine. Below are some example distros.
|
||||
|
||||
@@ -26,31 +21,25 @@ Install QEMU-KVM
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
swupd bundle-add kvm-host
|
||||
sudo swupd bundle-add kvm-host
|
||||
|
||||
* |CL| option: Add this bundle to lanch the GUI upon boot.
|
||||
* On Ubuntu\* 18.04 LTS Desktop:
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
swupd bundle-add desktop-autostart
|
||||
sudo apt-get install qemu-kvm
|
||||
|
||||
* On Ubuntu\* 16.04 LTS Desktop:
|
||||
* On Mint\* 19.1 “Cinnamon” Desktop:
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
apt-get install qemu-kvm
|
||||
sudo apt-get install qemu-kvm
|
||||
|
||||
* On Mint\* 18.1 “Serena” Desktop:
|
||||
* On Fedora\* 30 Workstation:
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
apt-get install qemu-kvm
|
||||
|
||||
* On Fedora\* 25 Workstation:
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
dnf install qemu-kvm
|
||||
sudo dnf install qemu-kvm
|
||||
|
||||
Download and launch the virtual machine
|
||||
***************************************
|
||||
@@ -67,26 +56,26 @@ Download and launch the virtual machine
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
unxz clear-<version>-kvm.img.xz
|
||||
unxz -v clear-<version>-kvm.img.xz
|
||||
|
||||
#. Download the `OVMF file`_ file that provides UEFI support for
|
||||
virtual machines.
|
||||
#. Download the 3 OVMF files (`OVMF.fd`, `OVMF_CODE.fd`, `OVMF_VARS.fd`) that
|
||||
provides UEFI support for virtual machines.
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
curl -O https://cdn.download.clearlinux.org/image/OVMF.fd
|
||||
|
||||
#. Copy :file:`OVMF.fd` to the working directory, as shown below.
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
cp OVMF.fd <working_directory>
|
||||
curl -O https://cdn.download.clearlinux.org/image/OVMF_CODE.fd
|
||||
curl -O https://cdn.download.clearlinux.org/image/OVMF_VARS.fd
|
||||
|
||||
.. note::
|
||||
|
||||
Replace <working_directory> with your own.
|
||||
The default OVMF files from |CL| may not work for some distro version(s).
|
||||
You may get an `ASSERT` in the `debug.log` file when starting the VM.
|
||||
If that is the case, use the distro-specific-OVMF files instead.
|
||||
For example, the |CL| OVMF files work for Ubuntu 18.04 LTS, but not for Ubuntu 19.04 LTS.
|
||||
Installing and using the OVMF files for Ubuntu 19.04 LTS resolved the `ASSERT` issue.
|
||||
|
||||
#. Download the sample `QEMU-KVM launcher`_ script from the
|
||||
#. Download the `QEMU-KVM launcher`_ script from the
|
||||
`image <https://cdn.download.clearlinux.org/image/>`_ directory. This script
|
||||
will launch the |CL| VM and provide console interaction within the same
|
||||
terminal emulator window.
|
||||
@@ -105,7 +94,7 @@ Download and launch the virtual machine
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
./start_qemu.sh clear-<version>-kvm.img
|
||||
sudo ./start_qemu.sh clear-<version>-kvm.img
|
||||
|
||||
#. Log in as ``root`` user and set a new password.
|
||||
|
||||
@@ -123,21 +112,28 @@ launched from, follow these steps.
|
||||
PermitRootLogin yes
|
||||
EOF
|
||||
|
||||
#. Start SSH server in the |CL| VM:
|
||||
#. Enable and start SSH server in the |CL| VM:
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
systemctl enable sshd
|
||||
systemctl start sshd
|
||||
|
||||
#. From the host, SSH into the |CL| VM. The port number ``10022`` is defined
|
||||
in the ``start_qemu.sh`` script.
|
||||
#. Determine the IP address of the host on which you will launch the VM.
|
||||
Substitute <ip-addr-of-kvm-host> in the next step with this information.
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
ssh -p 10022 root@localhost
|
||||
ip a
|
||||
|
||||
Add the GNOME Display Manager (GDM)
|
||||
***********************************
|
||||
#. SSH into the |CL| VM using the default port of `10022`:
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
ssh -p 10022 root@<ip-addr-of-kvm-host>
|
||||
|
||||
Optional: Add the GNOME Display Manager (GDM)
|
||||
*********************************************
|
||||
|
||||
To add :abbr:`GDM (GNOME Display Manager)` to the |CL| VM, follow these steps:
|
||||
|
||||
@@ -145,90 +141,90 @@ To add :abbr:`GDM (GNOME Display Manager)` to the |CL| VM, follow these steps:
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
shutdown now
|
||||
poweroff
|
||||
|
||||
#. Install a VNC viewer on the host machine. Below are some example distros.
|
||||
#. Install the Spice viewer on the local host or remote system. Below are some
|
||||
example distros.
|
||||
|
||||
* On Clear Linux:
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
swupd bundle-add tigervnc
|
||||
sudo swupd bundle-add virt-viewer
|
||||
|
||||
* On Ubuntu\* 16.04 LTS Desktop:
|
||||
* On Ubuntu\* 18.04 LTS Desktop:
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
apt-get install vncviewer
|
||||
sudo apt-get install virt-viewer
|
||||
|
||||
* On Mint\* 18.1 “Serena” Desktop:
|
||||
* On Mint\* 19.1 “Cinnamon” Desktop:
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
apt-get install vncviewer
|
||||
sudo apt-get install virt-viewer
|
||||
|
||||
* On Fedora\* 25 Workstation:
|
||||
* On Fedora\* 30 Workstation:
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
dnf install tigervnc
|
||||
sudo dnf install virt-viewer
|
||||
|
||||
#. Modify the :file:`start_qemu.sh` script to increase memory (``-m``), add
|
||||
graphics driver (``-vga``), and add VNC (``-vnc``, ``-usb``, and
|
||||
``-device``) support.
|
||||
#. Modify the :file:`start_qemu.sh` script to increase memory (`-m`), add
|
||||
graphics driver (`-vga`), and add Spice (`-spice`, `-usb`, and
|
||||
`-device`) support.
|
||||
|
||||
.. code-block:: console
|
||||
|
||||
qemu-system-x86_64 \
|
||||
-enable-kvm \
|
||||
-bios OVMF.fd \
|
||||
${UEFI_BIOS} \
|
||||
-smp sockets=1,cpus=4,cores=2 -cpu host \
|
||||
-m 4096 \
|
||||
-vga qxl \
|
||||
-vnc :0 -nographic \
|
||||
-nographic \
|
||||
-spice port=5924,disable-ticketing \
|
||||
-usb \
|
||||
-device usb-tablet \
|
||||
-device usb-tablet,bus=usb-bus.0 \
|
||||
-drive file="$IMAGE",if=virtio,aio=threads,format=raw \
|
||||
-netdev user,id=mynet0,hostfwd=tcp::${VMN}0022-:22,hostfwd=tcp::${VMN}2375-:2375 \
|
||||
-device virtio-net-pci,netdev=mynet0 \
|
||||
-debugcon file:debug.log -global isa-debugcon.iobase=0x402 $@
|
||||
|
||||
#. Due to changes in the :file:`start_qemu.sh` script from the previous step,
|
||||
the UEFI :file:`NvVars`
|
||||
information for the previously-booted |CL| VM will need to be reset.
|
||||
using the same OVMF files will result in the VM not booting properly and
|
||||
you end up in the the UEFI shell. The easiest way to avoid this is to delete
|
||||
the OVMF files and restore the originals before relaunching the VM.
|
||||
|
||||
#. Relaunch the |CL| VM. The UEFI shell will appear.
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
./start_qemu.sh clear-<version>-kvm.img
|
||||
|
||||
#. At the UEFI shell, delete the :file:`NvVars` file:
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
Shell> del FS0:\NvVars
|
||||
|
||||
#. Exit out of the UEFI shell:
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
Shell> reset -s
|
||||
|
||||
#. Relaunch the |CL| VM:
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
./start_qemu.sh clear-<version>-kvm.img
|
||||
|
||||
#. From the host machine, open a new terminal emulator window and VNC into the
|
||||
|CL| VM:
|
||||
#. Increase the size of the VM by 10GB to accommodate the GDM installation:
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
vncviewer 0.0.0.0
|
||||
qemu-img resize -f raw clear-<version>-kvm.img +10G
|
||||
|
||||
#. Log in as ``root`` user into the |CL| VM.
|
||||
#. Relaunch the |CL| VM:
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
sudo ./start_qemu.sh clear-<version>-kvm.img
|
||||
|
||||
#. Determine the IP address of the host on which you will launch the VM.
|
||||
Substitute <ip-addr-of-kvm-host> in the next step with this information.
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
ip a
|
||||
|
||||
#. From the local host or remote system, open a new terminal emulator window
|
||||
and connect into the |CL| VM using the Spice viewer:
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
remote-viewer spice://<ip-address-of-kvm-host>:5924
|
||||
|
||||
#. Log in as `root` user into the |CL| VM.
|
||||
|
||||
#. Follow these steps to `resize the partition`_ of the virtual disk of the VM.
|
||||
|
||||
#. Add GDM to the |CL| VM:
|
||||
|
||||
@@ -236,19 +232,20 @@ To add :abbr:`GDM (GNOME Display Manager)` to the |CL| VM, follow these steps:
|
||||
|
||||
swupd bundle-add desktop-autostart
|
||||
|
||||
#. Reboot the |CL| VM to enable GDM:
|
||||
#. Reboot the |CL| VM to start GDM:
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
reboot
|
||||
|
||||
#. Go through GDM's out-of-box experience (OOBE).
|
||||
#. Go through the GDM out-of-box experience (OOBE).
|
||||
|
||||
#. The default aspect ratio of the GDM GUI for the |CL| VM is 4:3. To change
|
||||
it, use GDM's ``Displays`` setting tool (located at the top-right corner).
|
||||
#. The default aspect ratio of the GDM GUI for the |CL| VM is 4:3. To change
|
||||
it, use GDM's `Devices > Displays` setting tool (located at the top-right corner).
|
||||
|
||||
|
||||
.. _Intel® Virtualization Technology: https://www.intel.com/content/www/us/en/virtualization/virtualization-technology/intel-virtualization-technology.html
|
||||
.. _Intel®Virtualization Technology for Directed I/O: https://software.intel.com/en-us/articles/intel-virtualization-technology-for-directed-io-vt-d-enhancing-intel-platforms-for-efficient-virtualization-of-io-devices
|
||||
.. _QEMU-KVM launcher: https://cdn.download.clearlinux.org/image/start_qemu.sh
|
||||
.. _OVMF file: https://cdn.download.clearlinux.org/image/OVMF.fd
|
||||
.. _resize the partition: https://clearlinux.org/documentation/clear-linux/guides/maintenance/increase-virtual-disk-size#id8
|
||||
|
||||
@@ -79,7 +79,7 @@ The method below uses the command line tool *nmcli* to modify network
|
||||
connection.
|
||||
|
||||
|
||||
#. Identify the existing connection name.
|
||||
#. Identify the existing connection name:
|
||||
|
||||
.. code:: bash
|
||||
|
||||
@@ -114,7 +114,15 @@ connection.
|
||||
:file:`/etc/NetworkManager/system-connections/*.nmconnection`. can be edited
|
||||
directly.
|
||||
|
||||
#. Verify your static IP address details have been set
|
||||
|
||||
#. Restart the NetworkManager server to reload the DNS servers:
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
sudo systemctl restart NetworkManager
|
||||
|
||||
|
||||
#. Verify your static IP address details have been set:
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
@@ -167,7 +175,8 @@ created manually.
|
||||
|
||||
sudo systemctl restart systemd-networkd
|
||||
|
||||
#. Verify your static IP address details have been set
|
||||
|
||||
#. Verify your static IP address details have been set:
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
|
||||
+13
-11
@@ -23,12 +23,13 @@ Background
|
||||
|
||||
Software that is compiled and packaged for general usage in an operating
|
||||
system typically only contains components that are used to execute the
|
||||
program, like binaries and libraries. Extra developer data like the the
|
||||
actual source code and symbol information are separated and excluded for
|
||||
program, such as binaries and libraries. Extra developer data, such as the
|
||||
actual source code and symbol information, are separated and excluded for
|
||||
efficiency.
|
||||
|
||||
The debug information helps relate binary code to human readable source code
|
||||
lines and variables. Most of the time this auxiliary information is not needed
|
||||
lines and variables. Most of the time, this auxiliary information
|
||||
is not needed;
|
||||
however without it, debugging a program results in limited visibility.
|
||||
|
||||
|
||||
@@ -50,15 +51,16 @@ installed.
|
||||
clr-debug-info.
|
||||
|
||||
|
||||
#. Start a debugging session against a program using a debugger, like GDB.
|
||||
For example, to debug *gnome-control-center*:
|
||||
#. Start a debugging session against a program using a debugger, such as GDB.
|
||||
For example, to debug *gnome-control-center* execute the following
|
||||
command:
|
||||
|
||||
.. code:: bash
|
||||
|
||||
gdb /usr/bin/gnome-control-center
|
||||
|
||||
As the program is stepped through and debug information is needed, the
|
||||
clr_debug_daemon will obtain it in the background.
|
||||
As you step through the program and debug information is needed, the
|
||||
clr_debug_daemon obtains it in the background.
|
||||
|
||||
|
||||
Implementation
|
||||
@@ -68,20 +70,20 @@ The implementation of the |CL| debug system is open source and available on
|
||||
GitHub at: https://github.com/clearlinux/clr-debug-info/
|
||||
|
||||
.. figure:: figures/debug-diagram.png
|
||||
:scale: 75%
|
||||
:width: 400px
|
||||
:alt: Debug system communication flow
|
||||
|
||||
Figure 1: The communication flow of the |CL| debug system
|
||||
|
||||
The |CL| debug system implements a :abbr:`FUSE (filesystem in userspace)`
|
||||
filesystem mounted at :file:`/usr/lib/debug` and :file:`/usr/src/debug`. The
|
||||
FUSE filesystem starts automatically. Its status can be verified with
|
||||
FUSE filesystem starts automatically. You can verify its status by executing
|
||||
:command:`systemctl status clr_debug_fuse.service`.
|
||||
|
||||
The *clr_debug_daemon* is responsible for fetching the appropriate package
|
||||
debug content from the server and making it available for any debugging
|
||||
programs needed it. It is socket activated whenever a request to the local
|
||||
FUSE filesystem occurs. Its status can be verified with :command:`systemctl
|
||||
programs that need it. It is socket activated whenever a request to the local
|
||||
FUSE filesystem occurs. You can verify its status with :command:`systemctl
|
||||
status clr_debug_daemon.service`.
|
||||
|
||||
|
||||
|
||||
@@ -15,9 +15,9 @@ Description
|
||||
|
||||
:command:`swupd` has two main functions:
|
||||
|
||||
#. Manages software and replaces APT or YUM, by installing bundles
|
||||
#. Manage software and replace APT or YUM, by installing bundles
|
||||
rather than packages.
|
||||
#. Checks for system updates and installs them.
|
||||
#. Check for system updates and install them.
|
||||
|
||||
:command:`swupd` manages overlapping dependencies behind the scenes, ensuring
|
||||
that all software is compatible across the system. It can be used to verify
|
||||
@@ -25,27 +25,27 @@ the OS, clean cached files, and fix issues.
|
||||
|
||||
:ref:`Bundles <bundles>` contain everything needed to deliver a software
|
||||
capability. They are the smallest granularity component that is
|
||||
managed by |CL|. A bundle comes with all of its dependencies, rather than
|
||||
managed by |CL|. A bundle comes with all of its dependencies rather than
|
||||
downloading a cascade of package dependencies when installing a piece of
|
||||
software.
|
||||
|
||||
Versioning
|
||||
==========
|
||||
|
||||
Using package managers to keep track of software version compatibility or
|
||||
Using package managers to monitor software version compatibility or
|
||||
compare multiple systems on many Linux distributions can be cumbersome.
|
||||
|
||||
With |CL| :command:`swupd`, versioning happens at the individual file-level.
|
||||
With |CL| :command:`swupd`, versioning happens at the individual file level.
|
||||
This means |CL| generates an entirely new OS version with any set of software
|
||||
changes to the system (including software downgrades or removals). This
|
||||
changes to the system, including software downgrades or removals. This
|
||||
rolling release versioning model is similar to :command:`git` internal version
|
||||
tracking, where any of the individual file commits are tracked and move the
|
||||
pointer forward when changed.
|
||||
|
||||
A number that represents the **current** release of the OS describes the
|
||||
versions of all the software on the OS. Each build is composed of a specific
|
||||
set of bundles made from a particular version of packages. This matters on a
|
||||
daily basis to system administrators, who need to determine which of their
|
||||
set of bundles made from a particular version of packages. On a daily basis,
|
||||
this matters to system administrators who need to determine which of their
|
||||
systems do not have the latest security fixes, or which combinations of
|
||||
software have been tested. Every release of the same number is guaranteed to
|
||||
contain the same versions of software, so there's no ambiguity between two
|
||||
@@ -59,8 +59,8 @@ checks for updates against a version server. The content server provides the
|
||||
file and metadata content for all versions and can be the same as the
|
||||
version server. The content url server provides metadata in the form of
|
||||
*manifests*, which list and describe file contents, symlinks,
|
||||
and directories. Additionally, the actual content is
|
||||
provided to clients in the form of archive files.
|
||||
and directories. Additionally, the actual content is provided to clients
|
||||
in the form of archive files.
|
||||
|
||||
Software updates with |CL| are also efficient. Unlike package-based
|
||||
distributions, :command:`swupd` only updates files that have changed, rather
|
||||
@@ -121,7 +121,7 @@ a system is obtained by one of the following methods:
|
||||
Bundle search
|
||||
=============
|
||||
|
||||
:command:`swupd` search downloads manifest data and searches for
|
||||
:command:`swupd` searches download manifest data for
|
||||
bundles that match the term. Enter only one term, or hyphenated term, per
|
||||
search. Use the command :command:`man swupd` to learn more.
|
||||
|
||||
@@ -187,7 +187,7 @@ on demand.
|
||||
|
||||
.. _swupd-guide-example-install-bundle:
|
||||
|
||||
Example 2: Find and install Kata\* Containers
|
||||
Example 2: Find and install Kata Containers\*
|
||||
=============================================
|
||||
|
||||
Kata Containers is a popular container implementation. Unlike other
|
||||
@@ -199,7 +199,7 @@ improved security.
|
||||
`one bundle`_ to use `Kata Containers`_: `containers-virt`, despite a
|
||||
number of dependencies. Also, check out our tutorial: :ref:`kata`.
|
||||
|
||||
#. Find the right bundle.
|
||||
#. Find the correct bundle.
|
||||
|
||||
To return all possible matches for the search string, enter
|
||||
:command:`swupd search`, followed by 'kata':
|
||||
|
||||
@@ -53,8 +53,9 @@ Stack features
|
||||
|
||||
.. note::
|
||||
|
||||
Performance test results for the Deep Learning Reference Stack were
|
||||
obtained using `runc` as the runtime.
|
||||
The Deep Learning Reference Stack is a collective work, and each piece of software within the work has its own license. Please see the `terms of use`_ for more details about licensing and usage of the Deep Learning Reference Stack.
|
||||
|
||||
|
||||
|
||||
Prerequisites
|
||||
=============
|
||||
@@ -107,6 +108,12 @@ For multi-node testing, replicate these steps for each node. These steps
|
||||
provide a template to run other benchmarks, provided that they can invoke
|
||||
TensorFlow.
|
||||
|
||||
.. note::
|
||||
|
||||
Performance test results for the Deep Learning Reference Stack and for this tutorial were
|
||||
obtained using `runc` as the runtime.
|
||||
|
||||
|
||||
#. Download either the `Eigen`_ or the `Intel MKL-DNN`_ Docker image
|
||||
from `Docker Hub`_.
|
||||
|
||||
@@ -243,7 +250,7 @@ Images
|
||||
|
||||
You must add `launcher.py`_ to the Docker image to include the Deep
|
||||
Learning Reference Stack and put the benchmarks repo in the correct
|
||||
location. Note that this tutorial uses Kubeflow v0.4.0, and cannot guarantee results if you use a different version.
|
||||
location. Note that this tutorial uses Kubeflow v0.4.0, and cannot guarantee results if you use a different version.
|
||||
|
||||
From the Docker image, run the following:
|
||||
|
||||
@@ -495,6 +502,8 @@ Related topics
|
||||
|
||||
.. _launcher.py: https://github.com/clearlinux/dockerfiles/tree/master/stacks/dlrs/kubeflow
|
||||
|
||||
.. _terms of use: https://clearlinux.org/stacks/deep-learning/terms-of-use
|
||||
|
||||
.. _Release notes on Github\*: https://github.com/clearlinux/dockerfiles/blob/master/stacks/dlrs/releasenote.md
|
||||
|
||||
.. _Intel® quantization tools: https://github.com/IntelAI/tools/blob/master/tensorflow_quantization/README.md#quantization-tools
|
||||
|
||||
+53
-15
@@ -14,10 +14,15 @@ automatically at system boot if a compatible card is detected.
|
||||
These instructions show how to use the proprietary NVIDIA drivers which
|
||||
require a manual installation.
|
||||
|
||||
.. note::
|
||||
.. warning::
|
||||
|
||||
Software installed outside of :ref:`swupd <swupd-about>` is not updated with |CL|
|
||||
updates and must be updated and maintained manually.
|
||||
Software installed outside of :ref:`swupd <swupd-about>` is not updated
|
||||
with |CL| updates and must be updated and maintained manually.
|
||||
|
||||
For example, the file :file:`/usr/lib/libGL.so` conflicts between the one
|
||||
provided by the mesa package in |CL| and the one NVIDIA provides. If a |CL|
|
||||
update overwrites these files, a reinstallation of the NVIDIA driver might
|
||||
be required.
|
||||
|
||||
|
||||
|
||||
@@ -107,17 +112,20 @@ needs to be disabled before installation can continue.
|
||||
|
||||
|
||||
|
||||
Configure the Dynamic Linker
|
||||
============================
|
||||
|
||||
Configure Alternative Software Paths
|
||||
====================================
|
||||
|
||||
The NVIDIA installer will be directed to install files under
|
||||
:file:`/opt/nvidia` as much as possible to keep its contents isolated from the
|
||||
rest of the |CL| system files under :file:`/usr`. The dynamic linker will
|
||||
need to be configured to use the NVIDIA-provided libraries.
|
||||
rest of the |CL| system files under :file:`/usr`. The dynamic linker and X
|
||||
server will need to be configured to use the content under
|
||||
:file:`/opt/nvidia`.
|
||||
|
||||
|
||||
#. Configure the dynamic linker to look for and cache shared libraries under
|
||||
:file:`/opt/nvidia/lib` and :file:`/opt/nvidia/lib32`.
|
||||
:file:`/opt/nvidia/lib` and :file:`/opt/nvidia/lib32` in addition to the
|
||||
default paths.
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
@@ -125,7 +133,26 @@ need to be configured to use the NVIDIA-provided libraries.
|
||||
|
||||
sudo mkdir /etc/ld.so.conf.d
|
||||
printf "/opt/nvidia/lib \n/opt/nvidia/lib32 \n" | sudo tee --append /etc/ld.so.conf.d/nvidia.conf
|
||||
|
||||
#. Reload the dynamic linker run-time bindings and library cache.
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
sudo ldconfig
|
||||
|
||||
#. Create a Xorg configuration file to search for modules under
|
||||
:file:`/opt/nvidia` in addition to the default path.
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
sudo mkdir -p /etc/X11/xorg.conf.d/
|
||||
|
||||
sudo tee /etc/X11/xorg.conf.d/nvidia-files-opt.conf > /dev/null <<'EOF'
|
||||
Section "Files"
|
||||
ModulePath "/usr/lib64/xorg/modules"
|
||||
ModulePath "/opt/nvidia/lib64/xorg/modules"
|
||||
EndSection
|
||||
EOF
|
||||
|
||||
|
||||
Install the NVIDIA Drivers
|
||||
@@ -152,17 +179,22 @@ Install the NVIDIA Drivers
|
||||
--utility-prefix=/opt/nvidia \
|
||||
--opengl-prefix=/opt/nvidia \
|
||||
--compat32-prefix=/opt/nvidia \
|
||||
--compat32-libdir=lib32 \
|
||||
--compat32-libdir=lib32 \
|
||||
--x-prefix=/opt/nvidia \
|
||||
--x-module-path=/opt/nvidia/lib64/xorg/modules \
|
||||
--x-library-path=/opt/nvidia/lib64 \
|
||||
--x-sysconfig-path=/etc/X11/xorg.conf.d \
|
||||
--documentation-prefix=/opt/nvidia \
|
||||
--application-profile-path=/etc/nvidia \
|
||||
--no-precompiled-interface \
|
||||
--no-nvidia-modprobe \
|
||||
--no-distro-scripts \
|
||||
--force-libglx-indirect \
|
||||
--glvnd-egl-config-path=/etc/glvnd/egl_vendor.d \
|
||||
--egl-external-platform-config-path=/etc/egl/egl_external_platform.d \
|
||||
--dkms \
|
||||
--silent
|
||||
|
||||
|
||||
#. The graphical interface may automatically start after the NVIDIA driver
|
||||
is loaded. Return to the working terminal and log back in if necessary.
|
||||
|
||||
@@ -171,7 +203,7 @@ Install the NVIDIA Drivers
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
lsmod | grep ^nvidia
|
||||
lsmod | grep ^nvidia
|
||||
|
||||
|
||||
#. Run a |CL| system verification to restore files that the NVIDIA installer
|
||||
@@ -185,7 +217,8 @@ Install the NVIDIA Drivers
|
||||
|
||||
The NVIDIA software places some files under the :file:`/usr` subdirectory
|
||||
which are not managed by |CL| and conflict with the |CL| stateless design.
|
||||
Although a limited version of :command:`swupd repair` is ran above,
|
||||
|
||||
Although a limited version of :command:`swupd repair` is run above,
|
||||
other uses of the :command:`swupd repair` command should be avoided
|
||||
with the proprietary NVIDIA drivers installed.
|
||||
|
||||
@@ -198,7 +231,7 @@ The proprietary NVIDIA drivers are installed manually outside of :ref:`swupd
|
||||
|
||||
Updating the NVIDIA drivers follows the same steps as initial installation,
|
||||
however the desktop environment must first be stopped so that the drivers are
|
||||
not in use.
|
||||
not in use.
|
||||
|
||||
#. Follow the steps in `Download the NVIDIA Drivers for Linux`_ section to get
|
||||
the latest NVIDIA drivers.
|
||||
@@ -241,14 +274,19 @@ Uninstalling the NVIDIA Drivers
|
||||
The NVIDIA drivers and associated software can be uninstalled and nouveau
|
||||
driver restored by:
|
||||
|
||||
#. Remove the previously created file :file:`/etc/modprobe.d` that
|
||||
prevents nouveau from loading.
|
||||
#. Remove the :file:`modprobe.d` file that prevents nouveau from loading.
|
||||
|
||||
.. code-block:: bash
|
||||
|
||||
sudo rm /etc/modprobe.d/disable-nouveau.conf
|
||||
|
||||
|
||||
#. Remove the :file`xorg.conf.d` file that adds a search path for X modules.
|
||||
|
||||
.. code:: bash
|
||||
|
||||
sudo rm /etc/X11/xorg.conf.d/nvidia-files-opt.conf
|
||||
|
||||
#. Run the :command:`sudo /opt/nvidia/bin/nvidia-uninstall`
|
||||
|
||||
#. Follow the prompts on the screen and reboot the system.
|
||||
|
||||
Reference in New Issue
Block a user