Added step to change 'libvirt_type' when running compute on VM
Quite a few people initially try OpenStack by running each node on a virtual machine. In some cases, either the hypervisor and/or CPU lack support for hardware acceleration required by KVM on the compute node. I added a step including how to test for hardware acceleration support within the virtual machine and change 'libvirt_type' from KVM to QEMU if the test fails. Change-Id: I519cdff8c2c1b6a8d1d2b9d5181b205362651b85 backport: havana Closes-Bug: #1264175
This commit is contained in:
parent
1ceb035ee4
commit
a37b8d0c18
@ -180,7 +180,33 @@ novncproxy_base_url=http://<replaceable>controller</replaceable>:6080/vnc_auto.h
|
||||
...
|
||||
glance_host=<replaceable>controller</replaceable></programlisting>
|
||||
</step>
|
||||
|
||||
<step>
|
||||
<para>If you install Compute on a virtual machine for testing
|
||||
purposes, you must determine whether your hypervisor and/or CPU
|
||||
support nested hardware acceleration using the following
|
||||
command:</para>
|
||||
<screen><prompt>#</prompt> <userinput>egrep -c '(vmx|svm)' /proc/cpuinfo</userinput></screen>
|
||||
<para>If this command returns a value of
|
||||
<emphasis>one or greater</emphasis>, your hypervisor and/or CPU
|
||||
support nested hardware acceleration which requires no
|
||||
additional configuration.</para>
|
||||
<para os="ubuntu;debian">If this command returns a value of
|
||||
<emphasis>zero</emphasis>, your hypervisor and/or CPU do not
|
||||
support nested hardware acceleration and
|
||||
<literal>libvirt</literal> must use QEMU instead of KVM. Edit
|
||||
the <literal>[DEFAULT]</literal> section in the
|
||||
<filename>/etc/nova/nova-compute.conf</filename> file to modify
|
||||
this key:</para>
|
||||
<programlisting os="ubuntu;debian" language="ini">[DEFAULT]
|
||||
...
|
||||
libvirt_type = qemu</programlisting>
|
||||
<para os="rhel;centos;fedora;sles;opensuse">If this command returns
|
||||
a value of <emphasis>zero</emphasis>, your hypervisor and/or CPU
|
||||
do not support nested hardware acceleration and
|
||||
<literal>libvirt</literal> must use QEMU instead of KVM.
|
||||
Configure <literal>libvirt</literal> to use QEMU:</para>
|
||||
<screen os="rhel;centos;fedora;sles;opensuse"><prompt>#</prompt> <userinput>openstack-config --set /etc/nova/nova.conf DEFAULT libvirt_type qemu</userinput></screen>
|
||||
</step>
|
||||
<step>
|
||||
<para>Edit the <filename>/etc/nova/api-paste.ini</filename>
|
||||
file to add the credentials to the
|
||||
|
Loading…
Reference in New Issue
Block a user