openstack-manuals/doc/image-guide/section_centos-example.xml
Ramaraja cc381368b0 Update with correct script location for CentOS
Closes-Bug: #1412482

Change-Id: Ifd1819a067c9a4274baa5dc531411ac298052b08
2015-01-28 18:24:39 +00:00

332 lines
18 KiB
XML

<?xml version="1.0" encoding="UTF-8"?>
<section xmlns="http://docbook.org/ns/docbook" xmlns:xi="http://www.w3.org/2001/XInclude"
xmlns:xlink="http://www.w3.org/1999/xlink" version="5.0" xml:id="centos-image">
<title>Example: CentOS image</title>
<para>This example shows you how to install a CentOS image and focuses mainly on CentOS 6.4.
Because the CentOS installation process might differ across versions, the installation steps
might differ if you use a different version of CentOS.</para>
<procedure>
<title>Download a CentOS install ISO</title>
<step>
<para>Navigate to the <link
xlink:href="http://www.centos.org/download/mirrors/">CentOS
mirrors</link> page.</para>
</step>
<step>
<para>Click one of the <literal>HTTP</literal> links in the right-hand column next to
one of the mirrors.</para>
</step>
<step>
<para>Click the folder link of the CentOS version that you want to use. For example,
<literal>6.4/</literal>.</para>
</step>
<step>
<para>Click the <literal>isos/</literal> folder link.</para>
</step>
<step>
<para>Click the <literal>x86_64/</literal> folder link for 64-bit images.</para>
</step>
<step>
<para>Click the netinstall ISO image that you want to download. For example,
<filename>CentOS-6.4-x86_64-netinstall.iso</filename> is a good choice because
it is a smaller image that downloads missing packages from the Internet during
installation.</para>
</step>
</procedure>
<simplesect>
<title>Start the installation process</title>
<para>Start the installation process using either <command>virt-manager</command> or
<command>virt-install</command> as described in the previous section. If you use
<command>virt-install</command>, do not forget to connect your VNC client to the
virtual machine.</para>
<para>Assume that:</para>
<itemizedlist>
<listitem>
<para>The name of your virtual machine image is <literal>centos-6.4</literal>;
you need this name when you use <command>virsh</command> commands to manipulate the
state of the image.</para>
</listitem>
<listitem>
<para>You saved the netinstall ISO image to the <filename>/data/isos</filename> directory.</para>
</listitem>
</itemizedlist>
<para>If you use <command>virt-install</command>, the commands should look something like
this:</para>
<screen><prompt>#</prompt> <userinput>qemu-img create -f qcow2 /tmp/centos-6.4.qcow2 10G</userinput>
<prompt>#</prompt> <userinput>virt-install --virt-type kvm --name centos-6.4 --ram 1024 \
--disk /tmp/centos-6.4.qcow2,format=qcow2 \
--network network=default \
--graphics vnc,listen=0.0.0.0 --noautoconsole \
--os-type=linux --os-variant=rhel6 \
--extra-args="console=tty0 console=ttyS0,115200n8 serial" \
--location=/data/isos/CentOS-6.4-x86_64-netinstall.iso</userinput></screen>
</simplesect>
<simplesect>
<title>Step through the installation</title>
<para>At the initial Installer boot menu, choose the <guilabel>Install or upgrade an
existing system</guilabel> option. Step through the installation prompts. Accept the
defaults.</para>
<mediaobject>
<imageobject role="fo">
<imagedata fileref="figures/centos-install.png" format="PNG" scale="60"/>
</imageobject>
<imageobject role="html">
<imagedata fileref="figures/centos-install.png" format="PNG"/>
</imageobject>
</mediaobject>
</simplesect>
<simplesect>
<title>Configure TCP/IP</title>
<para>The default TCP/IP settings are fine. In particular, ensure that Enable IPv4 support
is enabled with DHCP, which is the default.</para>
<mediaobject>
<imageobject>
<imagedata fileref="figures/centos-tcpip.png" format="PNG" contentwidth="6in"/>
</imageobject>
</mediaobject>
</simplesect>
<simplesect>
<title>Point the installer to a CentOS web server</title>
<para>Choose URL as the installation method.</para>
<mediaobject>
<imageobject>
<imagedata fileref="figures/install-method.png" format="PNG" contentwidth="6in"/>
</imageobject>
</mediaobject>
<para>Depending on the version of CentOS, the net installer requires that the user specify
either a URL or the web site and a CentOS directory that corresponds to one of the
CentOS mirrors. If the installer asks for a single URL, a valid URL might be
<literal>http://mirror.umd.edu/centos/6/os/x86_64</literal>.</para>
<note>
<para>Consider using other mirrors as an alternative to
<literal>mirror.umd.edu</literal>.</para>
</note>
<mediaobject>
<imageobject>
<imagedata fileref="figures/url-setup.png" format="PNG" contentwidth="6in"/>
</imageobject>
</mediaobject>
<para>If the installer asks for web site name and CentOS directory separately, you might
enter:</para>
<itemizedlist>
<listitem>
<para>Web site name: <literal>mirror.umd.edu</literal></para>
</listitem>
<listitem>
<para>CentOS directory: <literal>centos/6/os/x86_64</literal></para>
</listitem>
</itemizedlist>
<para>See <link xlink:href="http://www.centos.org/download/mirrors/"
>CentOS mirror page</link> to get a full list of mirrors, click on the "HTTP" link
of a mirror to retrieve the web site name of a mirror.</para>
</simplesect>
<simplesect>
<title>Storage devices</title>
<para>If prompted about which type of devices your installation uses, choose <guilabel>Basic
Storage Devices</guilabel>.</para>
</simplesect>
<simplesect>
<title>Hostname</title>
<para>The installer may ask you to choose a host name. The default
(<literal>localhost.localdomain</literal>) is fine. You install the <systemitem
class="service">cloud-init</systemitem> package later, which sets the host name on
boot when a new instance is provisioned using this image.</para>
</simplesect>
<simplesect>
<title>Partition the disks</title>
<para>There are different options for partitioning the disks. The default installation uses
LVM partitions, and creates three partitions (<filename>/boot</filename>,
<filename>/</filename>, swap), which works fine. Alternatively, you might want to
create a single ext4 partition that is mounted to "<literal>/</literal>", which also
works fine.</para>
<para>If unsure, use the default partition scheme for the installer because no scheme is
better than another.</para>
</simplesect>
<simplesect>
<title>Step through the installation</title>
<para>Step through the installation, using the default options. The simplest thing to do is
to choose the "Basic Server" install (may be called "Server" install on older versions
of CentOS), which installs an SSH server.</para>
</simplesect>
<simplesect>
<title>Detach the CD-ROM and reboot</title>
<para>After the install completes, the <guilabel>Congratulations, your CentOS installation
is complete</guilabel> screen appears.</para>
<mediaobject>
<imageobject>
<imagedata fileref="figures/centos-complete.png" format="PNG" contentwidth="6in"/>
</imageobject>
</mediaobject>
<para>To eject a disk by using the <command>virsh</command> command, libvirt requires that
you attach an empty disk at the same target that the CDROM was previously attached,
which should be <literal>hdc</literal>. You can confirm the appropriate target using the
<command>dom dumpxml <replaceable>vm-image</replaceable></command> command.</para>
<screen><prompt>#</prompt> <userinput>virsh dumpxml centos-6.4</userinput>
<computeroutput>&lt;domain type='kvm'>
&lt;name>centos-6.4&lt;/name>
...
&lt;disk type='block' device='cdrom'>
&lt;driver name='qemu' type='raw'/>
&lt;target dev='hdc' bus='ide'/>
&lt;readonly/>
&lt;address type='drive' controller='0' bus='1' target='0' unit='0'/>
&lt;/disk>
...
&lt;/domain>
</computeroutput></screen>
<para>Run the following commands from the host to eject the disk and reboot using virsh, as
root. If you are using virt-manager, the commands below will work, but you can also use
the GUI to detach and reboot it by manually stopping and starting.</para>
<screen><prompt>#</prompt> <userinput>virsh attach-disk --type cdrom --mode readonly centos-6.4 "" hdc</userinput>
<prompt>#</prompt> <userinput>virsh destroy centos-6.4</userinput>
<prompt>#</prompt> <userinput>virsh start centos-6.4</userinput></screen>
</simplesect>
<simplesect>
<title>Log in to newly created image</title>
<para>When you boot for the first time after installation, you might be prompted about
authentication tools. Select <guilabel>Exit</guilabel>. Then, log in as root.</para>
</simplesect>
<simplesect>
<title>Install the ACPI service</title>
<para>To enable the hypervisor to reboot or shutdown an instance, you
must install and run the <systemitem
class="service">acpid</systemitem> service on the guest
system.</para>
<para>Run the following commands inside the CentOS guest to install the
ACPI service and configure it to start when the system
boots:</para>
<screen><prompt>#</prompt> <userinput>yum install acpid</userinput>
<prompt>#</prompt> <userinput>chkconfig acpid on</userinput></screen>
</simplesect>
<simplesect>
<title>Configure to fetch metadata</title>
<para>An instance must interact with the metadata service to perform several tasks on start
up. For example, the instance must get the ssh public key and run the user data script.
To ensure that the instance performs these tasks, use one of these methods:</para>
<itemizedlist>
<listitem>
<para>Install a <systemitem class="service">cloud-init</systemitem> RPM, which is a
port of the Ubuntu <link xlink:href="https://launchpad.net/cloud-init"
>cloud-init</link> package. This is the recommended approach.</para>
</listitem>
<listitem>
<para>Modify <filename>/etc/rc.local</filename> to fetch desired information from
the metadata service, as described in the next section.</para>
</listitem>
</itemizedlist>
</simplesect>
<simplesect>
<title>Use cloud-init to fetch the public key</title>
<para>The <systemitem class="service">cloud-init</systemitem> package automatically fetches
the public key from the metadata server and places the key in an account. You can
install <systemitem class="service">cloud-init</systemitem> inside the CentOS guest by
adding the EPEL repo:</para>
<screen><prompt>#</prompt> <userinput>yum install http://download.fedoraproject.org/pub/epel/6/x86_64/epel-release-6-8.noarch.rpm</userinput>
<prompt>#</prompt> <userinput>yum install cloud-init</userinput></screen>
<para>The account varies by distribution. On Ubuntu-based virtual machines, the account is
called <literal>ubuntu</literal>. On Fedora-based virtual machines, the account is
called <literal>ec2-user</literal>.</para>
<para>You can change the name of the account used by <systemitem class="service"
>cloud-init</systemitem> by editing the <filename>/etc/cloud/cloud.cfg</filename>
file and adding a line with a different user. For example, to configure <systemitem
class="service">cloud-init</systemitem> to put the key in an account named
<literal>admin</literal>, add this line to the configuration file:</para>
<programlisting>user: admin</programlisting>
</simplesect>
<simplesect>
<title>Write a script to fetch the public key (if no cloud-init)</title>
<para>If you are not able to install the <systemitem class="service">cloud-init</systemitem>
package in your image, to fetch the ssh public key and add it to the root account, edit
the <filename>/etc/rc.d/rc.local</filename> file and add the following lines before the line
<literal>touch /var/lock/subsys/local</literal>:</para>
<programlisting language="bash">if [ ! -d /root/.ssh ]; then
mkdir -p /root/.ssh
chmod 700 /root/.ssh
fi
# Fetch public key using HTTP
ATTEMPTS=30
FAILED=0
while [ ! -f /root/.ssh/authorized_keys ]; do
curl -f http://169.254.169.254/latest/meta-data/public-keys/0/openssh-key \
> /tmp/metadata-key 2>/dev/null
if [ \$? -eq 0 ]; then
cat /tmp/metadata-key >> /root/.ssh/authorized_keys
chmod 0600 /root/.ssh/authorized_keys
restorecon /root/.ssh/authorized_keys
rm -f /tmp/metadata-key
echo "Successfully retrieved public key from instance metadata"
echo "*****************"
echo "AUTHORIZED KEYS"
echo "*****************"
cat /root/.ssh/authorized_keys
echo "*****************"
done</programlisting>
<note>
<para>Some VNC clients replace the colon (<literal>:</literal>) with a semicolon
(<literal>;</literal>) and the underscore (<literal>_</literal>) with a hyphen
(<literal>-</literal>). Make sure to specify <literal>http:</literal> and not
<literal>http;</literal>. Make sure to specify
<literal>authorized_keys</literal> and not
<literal>authorized-keys</literal>.</para>
</note>
<note>
<para>The previous script only gets the ssh public key from the metadata server. It does
not get user data, which is optional data that can be passed by the user when
requesting a new instance. User data is often used to run a custom script when an
instance boots.</para>
<para>As the OpenStack metadata service is compatible with version 2009-04-04 of the
Amazon EC2 metadata service, consult the Amazon EC2 documentation on <link
xlink:href="http://docs.amazonwebservices.com/AWSEC2/2009-04-04/UserGuide/AESDG-chapter-instancedata.html"
>Using Instance Metadata</link> for details on how to get user data.</para>
</note>
</simplesect>
<simplesect>
<title>Disable the zeroconf route</title>
<para>For the instance to access the metadata service, you must disable the default zeroconf
route:</para>
<screen><prompt>#</prompt> <userinput>echo "NOZEROCONF=yes" &gt;&gt; /etc/sysconfig/network</userinput></screen>
</simplesect>
<simplesect>
<title>Configure console</title>
<para>For the <command>nova console-log</command> command to work properly on CentOS
6.<replaceable>x</replaceable>, you might need to add the following lines to the
<filename>/boot/grub/menu.lst</filename> file:</para>
<programlisting>serial --unit=0 --speed=115200
terminal --timeout=10 console serial
# Edit the kernel line to add the console entries
kernel <replaceable>...</replaceable> console=tty0 console=ttyS0,115200n8</programlisting>
</simplesect>
<simplesect>
<title>Shut down the instance</title>
<para>From inside the instance, as root:</para>
<screen><prompt>#</prompt> <userinput>/sbin/shutdown -h now</userinput></screen>
</simplesect>
<simplesect>
<title>Clean up (remove MAC address details)</title>
<para>The operating system records the MAC address of the virtual Ethernet card in locations
such as <filename>/etc/sysconfig/network-scripts/ifcfg-eth0</filename> and
<filename>/etc/udev/rules.d/70-persistent-net.rules</filename> during the instance
process. However, each time the image boots up, the virtual Ethernet card will have a
different MAC address, so this information must be deleted from the configuration
file.</para>
<para>There is a utility called <command>virt-sysprep</command>, that performs various
cleanup tasks such as removing the MAC address references. It will clean up a virtual
machine image in place:</para>
<screen><prompt>#</prompt> <userinput>virt-sysprep -d centos-6.4</userinput></screen>
</simplesect>
<simplesect>
<title>Undefine the libvirt domain</title>
<para>Now that you can upload the image to the Image Service, you no longer need to have
this virtual machine image managed by libvirt. Use the <command>virsh undefine
<replaceable>vm-image</replaceable></command> command to inform libvirt:</para>
<screen><prompt>#</prompt> <userinput>virsh undefine centos-6.4</userinput></screen>
</simplesect>
<simplesect>
<title>Image is complete</title>
<para>The underlying image file that you created with <command>qemu-img create</command> is
ready to be uploaded. For example, you can upload the
<filename>/tmp/centos-6.4.qcow2</filename> image to the Image Service.</para>
</simplesect>
</section>