Attachment 7126 Details for Bug 16956 – Notes from previous test of xen

Notes from previous test of xen

xentest (text/plain), 3.27 KB, created by Dave Hodgins on 2015-10-14 20:29:55 CEST

(hide)

Description:

Filename:

MIME Type:

Creator: Dave Hodgins

Created: 2015-10-14 20:29:55 CEST

Size: 3.27 KB

patch

obsolete

>
>Ensure the host system is using kernel-server-latest.
>"uname -r" should show something like 
>3.8.13.3-server-2.mga3
>
>urpmi xen
>
>chkconfig --list|grep 5:on
>will show xencommons, xend, and xendomains are enabled. All three
>need to be disabled.
>
># systemctl disable xencommons.service
># systemctl disable xend.service
># systemctl disable xendomains.service
>
>The services xenconsoled and xenstored need to be enabled.
>
># systemctl enable xenconsoled.service
># systemctl enable xenstored.service
>
>On my amd system, having xend enabled will cause a hard reboot, when
>trying to boot into xen, forcing a full fsck on the the next boot,
>so be sure to correct which services are enabled before trying to
>boot into xen.
>
>Edit /boot/grub/menu.lst, copy the last boot stanza, which should
>be something like
>
>title server 3.8.13.3-2.mga3
>kernel (hd0,6)/boot/vmlinuz-3.8.13.3-server-2.mga3 BOOT_IMAGE=server_3.8.13.3-2.mga3 root=UUID=5699901f-6700-4d51-9a61-58a4bda7567b nokmsboot resume=LABEL="a3swap" vga=794
>root (hd0,6)
>initrd /boot/initrd-3.8.13.3-server-2.mga3.img
>
>Paste in a copy of the stanza, similar to the following
>
>title xen server 3.8.13.3-2.mga3
>kernel (hd0,7)/boot/xen.gz dom0_mem=4096MB
>module (hd0,7)/boot/vmlinuz-3.8.13.3-server-2.mga3 BOOT_IMAGE=server_3.8.13.3-2.mga3 root=UUID=f795db69-1357-4269-b17b-335430a8cff8 resume=UUID=9af64cdc-a611-47c3-821e-94a9fd60251e vga=788
>root (hd0,7)
>module /boot/initrd-3.8.13.3-server-2.mga3.img
>
>Note that the titile has been modified, to make it unique, the kernel
>line now loads xen.gz, and what was the kernel line, is now a module line,
>and the word initrd has also been replaced with the word module.
>
>Assuming eth0 is the network interface connecting to the internet,
># cd /etc/sysconfig/network-scripts/
># mv ifcfg-eth0 ifcfg-xenbr
>Edit ifcfg-xenbr
>Change the device from DEVICE=eth0 to DEVICE=xenbr, and add a line with
>TYPE=Bridge
>Note that the case is important in the type.
>
>Edit ifcfg-eth0, and past in
>DEVICE=eth0
>BRIDGE=xenbr
>ONBOOT=yes
>
>Edit /etc/shorewall/interfaces, and add a line with
>net	xenbr	detect
>
>Reboot, and select the xen stanza, from the grub menu.
>
>During the testing, 2 guest systems will be created. The first uses hvm
>(ada full virtualization, similar to VirtualBox), the second using pv,
>(aka partial virtualiztion), where the guest has direct access to more
>of the hosts hardware.
>
>The guest using hvm will be used to install a guest system, and then
>will be copied to create the starting point for the pv install.
>
>First create a sparse file, to contain the guests virtual hard drive.
># dd if=/dev/zero of=/opt/hvmtest.img count=1 bs=4M seek=4k
>Note that that the sparse file will start with using 4M, but as
>it's used, will grow, up to a max of 17GB, as shown by ...
># ls -lsh /opt/hvmtest.img 
>4.0M -rw-r--r-- 1 root root 17G Jun 27 21:28 /opt/hvmtest.img
>
>Create the config file used for installing the guest in the
>directory /etc/xen
>
>edit hvmtest, and save it with ...
>builder="hvm"
>name="hvmtest"
>vcpus=2
>memory=2048
>maxmem=4096
>vncviewer=1
>disk = [ 'file:/opt/hvmtest.img,sda,w', 'file:/s3/bkup/software/i3/boot-nonfree.iso,hdb:cdrom,r', ]
>vif = [ 'type=ioemu, mac=00:1f:5a:71:ae:37, bridge=xenbr', ] # choose a random mac
>boot="dc"
>
>Note that the mac address should be something unique for this guest, and
>use the correct name and path for an install iso.
>
>
>

Ensure the host system is using kernel-server-latest.
"uname -r" should show something like 
3.8.13.3-server-2.mga3

urpmi xen

chkconfig --list|grep 5:on
will show xencommons, xend, and xendomains are enabled. All three
need to be disabled.

# systemctl disable xencommons.service
# systemctl disable xend.service
# systemctl disable xendomains.service

The services xenconsoled and xenstored need to be enabled.

# systemctl enable xenconsoled.service
# systemctl enable xenstored.service

On my amd system, having xend enabled will cause a hard reboot, when
trying to boot into xen, forcing a full fsck on the the next boot,
so be sure to correct which services are enabled before trying to
boot into xen.

Edit /boot/grub/menu.lst, copy the last boot stanza, which should
be something like

title server 3.8.13.3-2.mga3
kernel (hd0,6)/boot/vmlinuz-3.8.13.3-server-2.mga3 BOOT_IMAGE=server_3.8.13.3-2.mga3 root=UUID=5699901f-6700-4d51-9a61-58a4bda7567b nokmsboot resume=LABEL="a3swap" vga=794
root (hd0,6)
initrd /boot/initrd-3.8.13.3-server-2.mga3.img

Paste in a copy of the stanza, similar to the following

title xen server 3.8.13.3-2.mga3
kernel (hd0,7)/boot/xen.gz dom0_mem=4096MB
module (hd0,7)/boot/vmlinuz-3.8.13.3-server-2.mga3 BOOT_IMAGE=server_3.8.13.3-2.mga3 root=UUID=f795db69-1357-4269-b17b-335430a8cff8 resume=UUID=9af64cdc-a611-47c3-821e-94a9fd60251e vga=788
root (hd0,7)
module /boot/initrd-3.8.13.3-server-2.mga3.img

Note that the titile has been modified, to make it unique, the kernel
line now loads xen.gz, and what was the kernel line, is now a module line,
and the word initrd has also been replaced with the word module.

Assuming eth0 is the network interface connecting to the internet,
# cd /etc/sysconfig/network-scripts/
# mv ifcfg-eth0 ifcfg-xenbr
Edit ifcfg-xenbr
Change the device from DEVICE=eth0 to DEVICE=xenbr, and add a line with
TYPE=Bridge
Note that the case is important in the type.

Edit ifcfg-eth0, and past in
DEVICE=eth0
BRIDGE=xenbr
ONBOOT=yes

Edit /etc/shorewall/interfaces, and add a line with
net	xenbr	detect

Reboot, and select the xen stanza, from the grub menu.

During the testing, 2 guest systems will be created. The first uses hvm
(ada full virtualization, similar to VirtualBox), the second using pv,
(aka partial virtualiztion), where the guest has direct access to more
of the hosts hardware.

The guest using hvm will be used to install a guest system, and then
will be copied to create the starting point for the pv install.

First create a sparse file, to contain the guests virtual hard drive.
# dd if=/dev/zero of=/opt/hvmtest.img count=1 bs=4M seek=4k
Note that that the sparse file will start with using 4M, but as
it's used, will grow, up to a max of 17GB, as shown by ...
# ls -lsh /opt/hvmtest.img 
4.0M -rw-r--r-- 1 root root 17G Jun 27 21:28 /opt/hvmtest.img

Create the config file used for installing the guest in the
directory /etc/xen

edit hvmtest, and save it with ...
builder="hvm"
name="hvmtest"
vcpus=2
memory=2048
maxmem=4096
vncviewer=1
disk = [ 'file:/opt/hvmtest.img,sda,w', 'file:/s3/bkup/software/i3/boot-nonfree.iso,hdb:cdrom,r', ]
vif = [ 'type=ioemu, mac=00:1f:5a:71:ae:37, bridge=xenbr', ] # choose a random mac
boot="dc"

Note that the mac address should be something unique for this guest, and
use the correct name and path for an install iso.

Actions: View

Attachments on bug 16956: 7126 | 7164