(Difference between revisions)
Jump to: navigation, search

Using EL6 hosts with Ovirt

(→‎Assumptions and Prerequisities)
Line 14: Line 14:
 
*** Server Platform Development
 
*** Server Platform Development
 
* An active webserver to host a package repository and the kickstart (EG: An EL6 or Fedora box with httpd loaded and running)
 
* An active webserver to host a package repository and the kickstart (EG: An EL6 or Fedora box with httpd loaded and running)
−
'''A Fair Warning, as it stands today getting an EL6 host to work and play nicely with ovirt-engine is not the faint of heart...'''
+
* '''At this point in time getting an EL6 host to work and play nicely with ovirt-engine is not trivial'''
  
 
== Building the Needed Packages ==
 
== Building the Needed Packages ==

Revision as of 16:51, 8 September 2012

How to use EL6 based hosts with Ovirt-Engine 3.1+

Contents

Assumptions and Prerequisities

  • An installed and running instance of ovirt-engine See: Installing_ovirt-engine_from_rpm
  • You will need to do the following on the system running your ovirt-engine to make it handle EL6 nodes instead of Fedora nodes:

For ovirt-engine 3.0 

 psql -U postgres -d engine -c "update vdc_options set option_value='rhel6.2.0' where option_name='EmulatedMachine' and version='3.0';"

For ovirt-engine 3.1 

 psql -U postgres -d engine -c "update vdc_options set option_value='rhel6.3.0' where option_name='EmulatedMachine' and version='3.1';"
  • A system or VM loaded with some flavor of EL6 EG: RHEL, CentOS, SL, OEL (Best practice is to use the one you intend to use for the node)
    • At minimum the following from the Development package group should be installed:
      • Development tools
      • Additional Development
      • Server Platform Development
  • An active webserver to host a package repository and the kickstart (EG: An EL6 or Fedora box with httpd loaded and running)
  • At this point in time getting an EL6 host to work and play nicely with ovirt-engine is not trivial

Building the Needed Packages

  • Some packages need to be built which will work with and are needed for VDSM
  • Some of these packages are not yet in EL6 or out of date for what upstream VDSM requires
  • Tend to stick with newer versions to match upstream VDSM

Building logrotate

  • Pull the latest logrotate source rpm from the Fedora 17 stable or updates repositories (use the newest)
  • Optionally if you don't mind living on the edge a bit you can pull the newest logrotate source RPM from the Rawhide repositories
  • PackageName: logrotate-<version>.fc<releasever>.src.rpm
  • The following packages are BuildRequires for logrotate
    • libselinux-devel
    • popt-devel
    • libacl-devel
  • To build logrotate: 
 rpmbuild --rebuild logrotate-<version>.fc<releasever>.src.rpm
  • Resulting RPMS:
    • logrotate-<version>.el6.x86_64.rpm
    • logrotate-debuginfo-<version>.el6.x86_64.rpm

Building sanlock

  • Pull the latest sanlock source rpm from the Fedora 17 stable or updates repositories (use the newest)
  • Optionally if you don't mind living on the edge a bit you can pull the newest sanlock source RPM from the Rawhide repositories
  • PackageName: sanlock-<version>.fc<releasever>.src.rpm
  • The following packages are BuildRequires for sanlock
    • libblkid-devel
    • libaio-devel
    • python python-devel
  • To build logrotate: 
 rpmbuild --rebuild sanlock-<version>.fc<releasever>.src.rpm
  • Resulting RPMS:
    • sanlock-<version>.el6.x86_64.rpm
    • sanlock-debuginfo-<version>.el6.x86_64.rpm
    • sanlock-devel-<version>.el6.x86_64.rpm
    • sanlock-lib-<version>.el6.x86_64.rpm
    • sanlock-python-<version>.el6.x86_64.rpm

Building Glusterfs

  • Pull the latest glusterfs source from Glusterfs LATEST
  • Source is a tar.gz and is named glusterfs-<version>.tar.gz
  • The following packages are BuildRequires for glusterfs
    • bison
    • flex
    • gcc
    • make
    • automake
    • libtool
    • ncurses-devel
    • readline-devel
    • openssl-devel
    • libxml2-devel
    • python-ctypes
    • python-devel
    • libibverbs-devel
  • To build glusterfs: 
 rpmbuild -tb glusterfs-<version>.tar.gz
  • Resulting RPMS:
    • glusterfs-<version>.el6.x86_64.rpm
    • glusterfs-debuginfo-<version>.el6.x86_64.rpm
    • glusterfs-devel-<version>.el6.x86_64.rpm
    • glusterfs-fuse-<version>.el6.x86_64.rpm
    • glusterfs-geo-replication-<version>.el6.x86_64.rpm
    • glusterfs-rdma-<version>.el6.x86_64.rpm
    • glusterfs-server-<version>.el6.x86_64.rpm

Building VDSM

  • Get the latest VDSM sources from GIT: 
 git clone http://gerrit.ovirt.org/p/vdsm
  • The following packages are BuildRequires for VDSM
    • python
    • python-devel
    • pyflakes
    • python-nose
    • python-pep8
    • python-ethtool
    • libvirt-python
    • sanlock-python >= 2.3
    • genisoimage
    • qemu-img (not in the vdsm.spec but seems to be needed by the vdsm-tests)
  • NOTE for the build requirement of sanlock-python you will need to install the following sanlock rpms you built prior:
    • sanlock-<version>.el6.x86_64.rpm
    • sanlock-lib-<version>.el6.x86_64.rpm
    • sanlock-python-<version>.el6.x86_64.rpm
  • Change directory into the resulting vdsm git source tree
  • Do the following to build VDSM

Run autogen.sh 

 ./autogen.sh

Run make 

 make

Make the rpms 

 make rpm
  • Resulting RPMS:
    • vdsm-<version>.el6.x86_64.rpm
    • vdsm-debuginfo-<version>.el6.x86_64.rpm
    • vdsm-python-<version>.el6.x86_64.rpm
    • vdsm-bootstrap-<version>.el6.noarch.rpm
    • vdsm-cli-<version>.el6.noarch.rpm
    • vdsm-debug-plugin-<version>.el6.noarch.rpm
    • vdsm-gluster-<version>.el6.noarch.rpm
    • vdsm-hook-faqemu-<version>.el6.noarch.rpm
    • vdsm-hook-qemucmdline-<version>.el6.noarch.rpm
    • vdsm-hook-vhostmd-<version>.el6.noarch.rpm
    • vdsm-reg-<version>.el6.noarch.rpm
    • vdsm-rest-<version>.el6.noarch.rpm
    • vdsm-tests-<version>.el6.noarch.rpm
    • vdsm-xmlrpc-<version>.el6.noarch.rpm

Deploying the RPMS via a yum repository

Hosting the repository

  • The easiest way to do this is to load a system with Fedora or EL6 and use an apache webserver to host the repository.
  • Setting up an apache webserver under EL6 is covered in depth here: Apache HTTP Server
  • Setting up an apache webserver under FC17 is covered in depth here: Apache HTTP Server

Creating the repository

  • Create a directory on the webserver at the desired location which will hold the rpms and the repository data.
    • Example for httpd on EL6: 
 mkdir /var/www/html/sl6v
  • Thus given the above location and assuming a webserver FQDN of ovirt.azeroth.net the URL for the repository would be: 
 http://ovirt.azeroth.net/sl6v/
  • You will now need to copy some of the rpms built previously into that hosted directory you created
  • Copy the following RPMS into the directory
    • logrotate-<version>.el6.x86_64.rpm
    • sanlock-<version>.el6.x86_64.rpm
    • sanlock-lib-<version>.el6.x86_64.rpm
    • sanlock-python-<version>.el6.x86_64.rpm
    • glusterfs-<version>.el6.x86_64.rpm
    • glusterfs-fuse-<version>.el6.x86_64.rpm
    • glusterfs-geo-replication-<version>.el6.x86_64.rpm
    • glusterfs-rdma-<version>.el6.x86_64.rpm
    • glusterfs-server-<version>.el6.x86_64.rpm
    • vdsm-<version>.el6.x86_64.rpm
    • vdsm-python-<version>.el6.x86_64.rpm
    • vdsm-cli-<version>.el6.noarch.rpm
    • vdsm-gluster-<version>.el6.noarch.rpm
    • vdsm-reg-<version>.el6.noarch.rpm
    • vdsm-xmlrpc-<version>.el6.noarch.rpm
  • Now we need to turn the directory into a yum repository to do this we will use createrepo
  • If createrepo is not installed on the system install it via: 
 yum -y install createrepo
  • Descend into the directory where you just copied the RPMS to and issue the following command: 
 createrepo -d .
  • You now have yum repository hosting the needed rpms built prior

Kickstarting Things

  • The best way to load a system to use as an ovirt VM host is to load it from a kickstart.
  • This allows for the following:
    • Automates the load and install process
    • Loads a minimal amount of packages (only what is needed and nothing more)
    • Reduces the install size and footprint
    • Saves time loading hosts
    • Makes it easier to reload hosts
  • Benefits over a stateless host
    • Can easily load or update drivers or custom 3rd party drivers
    • Configurations and data can be easily customized
    • Allows for loading of any additional software/packaged required for whatever reason
    • Easier to enable/disable services
    • Easier to enable/disable iptables or selinux
    • Easier to custom VDSM hooks
    • Easier to make customizations to networking
    • Easier to make customizations or changes via core libvirt where needed EG: virsh, libvirt networks, storage, etc
  • Kickstarting the load is best done via:
    • Booting from the DVD medi
    • PXE boot and install

Building the Kickstart

  • Building a kickstart for EL6 is covered more in depth here: Kickstart Installations
  • Included below is an example kickstart generated for using with Scientfic Linux 6.3
  • Things you might want to customize in the below kickstart:
    • Disk partitioning sizes and layout - the below load will wipe the partition tables and lay down a swap and "/" partition only
    • Enable the firewall if so desired
    • Enable selinux if so desired
    • Change the default system language
    • Change the default system keyboard layout
    • Change the timezone
  • Things you do want to change:
    • The #version=
      • Below we use SL so we call it SL6V for short for Scientific Linux 6.x Hypervisor
    • The root password
    • The repository definition for your distro
      • Below we are using SL so the repo definition for the base OS looks like:
Example: 
 repo --name="Scientific Linux 6.3 - x86_64" --baseurl="http://ftp.scientificlinux.org/linux/scientific/6.3/x86_64/os/"
  • The repository definition for the additional rpms
Example: 
 repo --name="sl6v" --baseurl="http://ovirt.azeroth.net/sl6v/"
  • The package <distroname>-release
    • Below we are using SL so it is sl-release change it whatever you are using EG: centos-release or redhat-release
  • Included in the below example kickstart is the package epel-release which SL includes yum repository definitions for EPEL
    • SL includes the EPEL repository defintions package
    • CentOS does not include it
    • RHEL does not include it
    • OEL does not include it
  • Thus in the case of the three EL6 distros which do not natively contain the EPEL repository definitions RPM you would need to add that RPM to the yum repository you created prior to hold the extra RPMS built above.
  • The EPEL repository definitions RPM can be found here: EPEL x86_64 Repo and is named: epel-release-<version>.noarch.rpm
  • Alternatively you can choose to remove the epel-release package from the package list in the example kickstart below.

Example EL6 Kickstart to load a minimal ovirt VM host: 

#platform=x86, AMD64, or Intel EM64T
#version=SL6V
# Firewall configuration
firewall --disabled
# Install OS instead of upgrade
install
# Use CDROM installation media
cdrom
# Root password
rootpw --plaintext <your password goes here>
# System authorization information
auth  --useshadow  --passalgo=sha512
# Use text mode install
text
# System keyboard
keyboard us
# System language
lang en_US
# SELinux configuration
selinux --disabled
# Do not configure the X Window System
skipx
# Installation logging level
logging --level=info
# Reboot after installation
reboot
# System timezone
timezone  America/Chicago
# System bootloader configuration
bootloader --append="elevator=deadline processor.max_cstate=1" --location=mbr
# Clear the Master Boot Record
zerombr
# Partition clearing information
clearpart --all --initlabel 
# Disk partitioning information
part swap --fstype="swap" --size=4096
part / --fstype="ext4" --grow --size=1

repo --name="Scientific Linux 6.3 - x86_64" --baseurl="http://ftp.scientificlinux.org/linux/scientific/6.3/x86_64/os/"
repo --name="your-repo-name" --baseurl="http://<url_to_the_server_hosting_your_rpms>/therpms/"

%packages --nobase
@ Core
ConsoleKit-libs
MAKEDEV
PyPAM
PyXML
SDL
acl
acpid
aic94xx-firmware
alsa-lib
attr
audit
audit-libs
audit-libs-python
augeas
augeas-libs
avahi-libs
basesystem
bash
bc
bfa-firmware
bind-libs
bind-utils
binutils
biosdevname
bridge-utils
btrfs-progs
busybox
bzip2
bzip2-libs
ca-certificates
celt051
chkconfig
cim-schema
coreutils
coreutils-libs
cpio
cracklib
cracklib-dicts
cracklib-python
cronie
cronie-anacron
crontabs
cryptsetup-luks
cryptsetup-luks-libs
curl
cyrus-sasl
cyrus-sasl-gssapi
cyrus-sasl-lib
cyrus-sasl-md5
dash
db4
db4-utils
dbus
dbus-glib
dbus-libs
dbus-python
device-mapper
device-mapper-event
device-mapper-event-libs
device-mapper-libs
device-mapper-multipath
device-mapper-multipath-libs
dhclient
dhcp-common
diffutils
dmidecode
dmraid
dnsmasq
dosfstools
dracut
dracut-fips
dracut-kernel
dracut-network
e2fsprogs
e2fsprogs-libs
ebtables
efibootmgr
eggdbus
eject
elfutils-libelf
elfutils-libs
epel-release
ethtool
expat
fence-agents
file
file-libs
filesystem
findutils
fipscheck
fipscheck-lib
flac
fuse
fuse-libs
gawk
gdb
gdbm
genisoimage
glib2
glibc
glibc-common
glusterfs
glusterfs-fuse
glusterfs-geo-replication
glusterfs-rdma
glusterfs-server
gmp
gnutls
gnutls-utils
gpgme
gpxe-roms-qemu
grep
groff
grub
grubby
gzip
hal
hal-info
hal-libs
hivex
hmaccalc
hwdata
info
initscripts
ipmitool
iproute
iptables
iptables-ipv6
iputils
irqbalance
iscsi-initiator-utils
kbd
kbd-misc
kernel
kernel-firmware
kexec-tools
keyutils
keyutils-libs
kpartx
krb5-libs
krb5-workstation
less
libICE
libSM
libX11
libX11-common
libXau
libXext
libXi
libXtst
libacl
libaio
libasyncns
libattr
libblkid
libcap
libcap-ng
libcgroup
libcmpiutil
libcom_err
libconfig
libcurl
libdrm
libedit
libevent
libffi
libgcc
libgcrypt
libgcrypt-devel
libgomp
libgpg-error
libgpg-error-devel
libgssglue
libgudev1
libguestfs
libguestfs-tools-c
libguestfs-winsupport
libibverbs
libidn
libjpeg
libmlx4
libnih
libnl
libogg
libpcap
libpciaccess
libselinux
libselinux-python
libselinux-utils
libsemanage
libsepol
libsndfile
libss
libssh2
libstdc++
libsysfs
libtasn1
libtirpc
libudev
libusb
libusb1
libuser
libuser-python
libutempter
libuuid
libvirt
libvirt-cim
libvirt-client
libvirt-lock-sanlock
libvirt-python
libvorbis
libxcb
libxml2
libxml2-python
libxslt
lm_sensors-libs
logrotate
lsof
lsscsi
ltrace
lua
lvm2
lvm2-libs
lzo
lzop
m2crypto
m4
man
mdadm
mesa-libGLU
mingetty
module-init-tools
nc
ncurses
ncurses-base
ncurses-libs
net-snmp
net-snmp-libs
net-snmp-utils
net-tools
netcf-libs
newt
newt-python
nfs-utils
nfs-utils-lib
nspr
nss
nss-softokn
nss-softokn-freebl
nss-sysinit
nss-tools
nss-util
ntp
ntpdate
numactl
numad
openldap
openssh
openssh-clients
openssh-server
openssl
pam
parted
passwd
patch
pciutils
pciutils-libs
pcre
perl-libs
pexpect
pinentry
pixman
pkgconfig
plymouth
plymouth-core-libs
plymouth-scripts
policycoreutils
policycoreutils-python
polkit
popt
procps
psmisc
pth
pulseaudio-libs
pyOpenSSL
pygobject2
pygpgme
pyparted
python
python-decorator
python-dmidecode
python-ethtool
python-gudev
python-hivex
python-iniparse
python-libguestfs
python-libs
python-pycurl
python-setuptools
python-simplejson
python-suds
python-urlgrabber
python-virtinst
qemu-img
qemu-kvm
qemu-kvm-tools
ql2100-firmware
ql2200-firmware
ql23xx-firmware
ql2400-firmware
ql2500-firmware
radvd
readline
redhat-logos
rootfiles
rpcbind
rpm
rpm-libs
rpm-python
rsync
rsyslog
sanlock
sanlock-lib
sanlock-python
sblim-sfcb
scrub
seabios
sed
selinux-policy
selinux-policy-targeted
setools-console
setup
sg3_utils
sg3_utils-libs
sgabios-bin
shadow-utils
slang
sl-release
sos
spice-server
sqlite
strace
sudo
sysfsutils
sysstat
system-config-keyboard-base
systemtap-runtime
sysvinit-tools
tar
tcp_wrappers-libs
tcpdump
telnet
traceroute
tree
tzdata
udev
unzip
upstart
usbredir
usbutils
ustr
util-linux-ng
vconfig
vgabios
vim-minimal
virt-what
virt-who
wget
which
xz
xz-libs
xz-lzma-compat
yajl
zlib
-NetworkManager
%end

Hosting the Kickstart

  • We can also use the same apache webserver we created for hosting the extra RPMS built prior to host the kickstart
  • Create another directory on the webserver to hold the kickstart:

Example for httpd on EL6: 

 mkdir /var/www/html/ks
  • Copy the kickstart file to the newly created directory

Example using sl6v.ks as the name of the kickstart file 

 cp sl6v.ks /var/www/html/ks/
  • Thus given the above location and assuming a webserver FQDN of ovirt.azeroth.net the URL for the kickstart would be: 
http://ovirt.azeroth.net/ks/sl6v.ks
  • This remotely accessible Kickstart can now be fed into a DVD or PXE install

Booting and installing via DVD

  • Booting and installing from the DVD media is the simplest method which works easily with most modern servers via their respective remote DVD media redirection capabilties:
    • Sun Oracle ILOM
    • Dell DRAC
    • HP ILO
  • EL6 Kickstart options are covered in depth here: Starting a Kickstart Installation
  • Assuming one of the above server media redirection methods attach the ISO of your choosen EL6 flavor to the remote re-direction console/media re-direct method.
  • Boot the server and choose to boot from the redirected media
  • Once at the initial grub boot menu of the install media highlight the "Install or upgrade an existing system " choice (this is the default in the case of all the EL6 distros)
  • Hit the tab key
  • Using the above URL to the kickstart you created and copied to your webserver add the following boot option 
 ks=http://url-to-your-webserver/kickstart-directory/name-of-you-kickstart-file
  • Next we need to tell it which network card to activate DHCP on and use to communicate on the network
  • Assuming the cable is plugged into the servers first ethernet port and that port will be "eth0" add the following boot option 
 ksdevice=eth0
  • Hit enter to start the install
  • If all goes well anaconda will find the kickstart and begin the automated install

Booting and installing via PXE

  • Setting up PXE boot and install under EL6 is covered in depth here: Setting Up an Installation Server
  • Setting up PXE boot and install under Fedora is covered in depth here: Manually configure a PXE server
  • EL6 Kickstart options are covered in depth here: Starting a Kickstart Installation
  • Assuming a PXE boot and install server has been setup per the above documentation
  • Assuming that a copy of your respective distro's pxeboot "vmlinuz" has been copied to a directory in /var/lib/tftpboot/ called "boot" EG: /var/lib/tftpboot/boot
  • Assuming that a copy of your respective distro's pxeboot "initrd" has been copied to a directory in /var/lib/tftpboot/ called "boot" EG: /var/lib/tftpboot/boot
  • Edit /var/lib/tftpboot/pxelinux.cfg/default and add the following: 
  label somelabel-goes-here
  kernel boot/vmlinuz
  append ks initrd=boot/initrd.img ramdisk_size=100000 ksdevice=eth0  repo=http://url-to-your-distros-base-install/ ks=http://url-to-your-webserver/kickstart-directory/name-of-you-kickstart-file
  • Given the above example eth0 is assumed as the ethernet port on the server which has an active link to the network

Where you are specifying that you copied the installation media or mirrored the install data for your choosen distro to: 

 repo=http://url-to-your-distros-base-install

Example Using SL: 

http://ftp.scientificlinux.org/linux/scientific/6.3/x86_64/os/

Where you specified that you copied the kickstart file created above: 

ks=http://url-to-your-webserver/kickstart-directory/name-of-you-kickstart-file

Example using the above kickstart: 

http://ovirt.azeroth.net/ks/sl6v.ks
  • Assuming all the configuration is correct boot your server via PXE
    • Usually F12 in most modern servers
    • OR Set the servers bootdev/order via the respective remote management method (ILOM, DRAC, ILO)
    • OR set the chassis bootdev to PXE via ipmitool to the respective remote BMC (ILOM, DRAC)
  • If all is well with the configuration the server will obtain an IP address from the ISC DHCP server you configured prior as part of the PXE server configuration
  • Once booted to the PXE server menu (assuming you have configured a menu, see EL6/Fedora documentation on this topic) type whatever label name you choose and hit return.
  • Alternatively you could have configured the PXE server to automatically kicked into your choosen label name (see EL6/Fedora documentation on this topic).
  • If all goes well the server will find the base repo fire up anaconda which will find the kickstart and begin the automated install

Setting up the Host

  • Provided all went well and you now have a cleanly loaded host via kickstart the system will now need some configuration.

Configure Networking

  • Need to configure networking on the system by editing/creating the following files
  • Below example configurations use:
  • A system named orgrimmar.azeroth.net (FQDN)
  • Uses eth0 and the according assigned physical ethernet port to talk to the ovirt-engine
  • Uses the IP/mask 192.168.1.1/24
  • Uses a default Gateway of 192.168.1.254
  • Uses the DNS domain of azeroth.net
  • Uses a DNS server at 192.168.1.100 for name resolution
  • Substitute your own values for the parameters in the below examples

Edit: /etc/hosts 

 127.0.0.1     localhost localhost.localdomain localhost4 localhost4.localdomain4
 ::1           localhost6 localhost6.localdomain6
 192.168.1.1    durotar durotar.azeroth.net

Edit: /etc/resolv.conf 

 search azeroth.net
 domain azeroth.net
 nameserver 192.168.1.100

Edit: /etc/sysconfig/network 

 NETWORKING=yes
 NETWORKING_IPV6=no
 HOSTNAME=durotar.azeroth.net
 GATEWAY=192.168.1.254

Edit /etc/sysconfig/network-scripts/ifcfg-eth0 

 DEVICE=eth0
 HWADDR=<HW MAC Address of eth0>
 ONBOOT=yes
 BOOTPROTO=none
 IPV6INIT=no
 BRIDGE=ovirtmgmt
 PEERDNS=no

Create: /etc/sysconfig/network-scripts/ifcfg-ovirtmgmt 

 DEVICE=ovirtmgmt
 ONBOOT=yes
 BOOTPROTO=none
 IPV6INIT=no
 TYPE=Bridge
 STP=off
 DELAY=0
 IPADDR=192.168.1.1
 PREFIX=24
 PEERDNS=no
  • Once networking is configured restart the network service 
 service network restart

Configure NTP

  • NTP needs to be configured and the host should talk to the same NTP server as the ovirt-engine which manages it.
  • The ovirt-engine and nodes need to be in sync with each other time wise
  • Also the ntpd service is required to be configured and running by VDSM
  • We use the example ntp server address of 192.168.101
  • Again substitute your own value for ntp server address

Edit: /etc/ntp.conf 

 driftfile /var/lib/ntp/drift
 includefile /etc/ntp/crypto/pw
 keys /etc/ntp/keys
 server 192.168.1.101

Optional Configurations

  • Change the ISCSI initiator name

Edit: /etc/iscsi/initiatorname.iscsi 

 InitiatorName=iqn.2012-08.net.azeroth:durotar
  • Perhaps more depending on how much you want to customize/modify

Add yum repository containing the extra rpms

  • We need to add a repository definition for the repo containing the extra rpms built prior
  • Uses the same values for the repo definition you defined in the kickstart
  • create a file in /etc/yum.repos.d/<your-repo-name>.repo

Example using the same examples in the above kickstart: 

[your-repo-name]
name=your-repo-name
baseurl=http://<url_to_the_server_hosting_your_rpms>/therpms/
enabled=1
gpgcheck=0

Actual Example with values (substitute in your own values): 

[sl6v]
name=Scientific Linux 6.3 Hypervisor
baseurl=http://ovirt.azeroth.net/sl6v/
enabled=1
gpgcheck=0

Bring things up to date

  • You will want to bring things up to date updates wise
  • In SL ensure both the security and fastbugs yum repositories are enabled
    • in /etc/yum.repos.d/sl.repo both "sl" and "sl-security" should be enabled (EG: enabled=1 for both)
    • in /etc/yum.repos.d/sl-other sl-fastbugs should be enabled (EG: enabled=1)
  • In CentOS both the base and updates repositories should be enabled
    • The CentOS fasttrack should also be enabled in /etc/yum.repos.d/CentOS-fasttrack.repo (EG: enabled=1)
  • In OEL ensure that ol6_latest and ol6_UEK_latest are enabled
    • in /etc/public-yum-ol6-.repo ol6_latest and ol6_UEK_latest should be enabled (EG: enabled=1 for both)
  • In RHEL the sytem should be subscribed and able to talk to RHN
  • Once the you ensure the proper yum repositories are enabled update the system via yum 
 yum -y update
  • Once the system is updated, reboot it
  • The system is now ready for assimilation into your ovirt-engine collective

Adding the Host to Ovirt

  • Assuming system has now been updated and configured

Disable un-needed services

  • Disable some services which will not be needed or will conflict:
  • libvirt-guests 
 chkconfig libvirt-guests off
  • glusterd
  • Currently used by Fedora hosts but with this setup could be used by your EL6 hosts too 
 chkconfig glusterd off

Optional services to enable/disable

  • Firewall can be optionally enabled or disabled
  • If it is enabled ports will need to be opened for SSH and VDSM (EG: 22 and 54321)

Enable the firewall: 

 chkconfig iptables on
 chkconfig ip6tables on

Disable the firewall: 

 chkconfig iptables off
 chkconfig ip6tables off
  • Enable or disable kernel core dumping

Enable kdump: 

 chkconfig kdump on

Disable kdump: 

 chkconfig kdump off
  • Any other services you may want enabled or disabled

Enable needed services

  • ntpd
  • iscsi
  • mutltipathd
  • Check the following services they should already be enabled and running
    • libvirtd
    • sanlock
    • sshd
    • wdmd

Configure SSH

  • Currently the deployutil.py in vdsm-bootstrap and vdsm-reg either fails to:
    • Copy the engine's SSH RSA key to EL6 based hosts authorized_keys file to EL6 based hosts authorized_keys file
    • Improperly copies the engines SSH RSA key to EL6 based hosts authorized_keys file to EL6 based hosts authorized_keys file
    • Or the deployutil.py getAuthKeysFile() function outright fails to run altogether when dealing with EL6 based hosts
  • Given the prior we can simply do the needful manually assuming we will be using vdsm-reg to add the host
  • ssh to the host from the ovirt-engine server and create the file /root/.ssh/authorized_keys
  • copy the contents of the file /etc/pki/ovirt-engine/keys/engine.ssh.key.txt to the file you just created on the host

Example; 

 mkdir /root/.ssh
 cat engine.ssh.key.txt >> /root/.ssh/authorized_keys
  • If adding the host via vds_bootstrap.py one can also just comment out the codeblock in the file (since we copied the key manually): 
 if not oDeploy.setSSHAccess(iurl, engine_ssh_key):
 logging.error('setSSHAccess test failed')
 return False

Install VDSM

  • Now you will need to install the VDSM rpms from your extra RPMS repository
  • Ensure you added (per the above) the repository definition for the repository you created which contains the VDSM RPMS you built and placed there previously

Install the neccesary VDSM RPMS: 

 yum -y install vdsm vdsm-python vdsm-cli vdsm-gluster vdsm-reg vdsm-xmlrpc
  • If this fails something is not right in your setup and you will want to trace back a few steps
  • Once the VDSM RPMS are installed you will want to turn off the vdsm-reg service (we will be manually invoking it) 
 chkconfig vdsm-reg off

Configure VDSM

  • Now that VDSM in installed we need to create a configuration for it
  • VDSM stores it's configuration file in /etc/vdsm/

Create: /etc/vdsm/vdsm.conf and add the following content to the file: 

 [vars]
 trust_store_path = /usr/local/etc/pki/vdsm
 ssl = true
 [addresses]
 management_port = 54321

Configure VDSM-reg

  • We will use vdsm-reg to assimilate the host into your ovirt-engine collective
  • We need to edit the configuration file for vdsm-reg stored in /etc/vdsm-reg/vdsm-reg.conf
  • In the below example configuration uses an ovirt-engine instance named ovirt.azeroth.net (FQDN)
  • You will need to substitute the FQDN of your ovirt-engine instance

Edit: /etc/vdsm-reg/vdsm-reg.conf 

  
[vars]
reg_req_interval = 5
vdsm_conf_file=/etc/vdsm/vdsm.conf
pidfile=/var/run/vdsm-reg.pid
logger_conf=/etc/vdsm-reg/logger.conf
vdc_host_name=ovirt.azeroth.net
vdc_host_port=443
vdc_reg_uri=/OvirtEngineWeb/register
#upgrade_iso_file=/data/updates/ovirt-node-image.iso
#upgrade_mount_point=/live
ticket=
  • Now that the ssh, vdsm, and vdsm-reg configurations are in place it's time to notify the ovirt-engine about our host

Start the vdsm-reg service 

 service vdsm-reg start
  • The vdsm-reg service will stop once it has successfully registered the host with your ovirt-engine instance
  • If registration was successful with your ovirt-engine instance you will see the host listed in the ovirt webadmin portal hosts tab and pending approval

Approve the host

  • Now that the host is registered go to the hosts tab in the ovirt webadmin portal
  • Click on the host and click approve
  • A dialog will pop up where you can configure
    • Choose a data center to assign the host to
    • Choose a cluster to assign the host to
    • Change or modify the name that ovirt refers to the host as
    • power management options (Note must have at least one other active host to do this)
    • the SPM priority of the host
  • If the host addition is successful the status of the host will change to "Up"
  • If all was successful the log entries should look something like: 
 <date>, <time> Host cluster Horde was updated by system 
 <date>, <time> Detected new Host orgrimmar Host state was set to Up. 
 <date>, <time> Host orgrimmar was successfully approved. 
 <date>, <time> Installing Host orgrimmar Step: RHEV_INSTALL. 
 <date>, <time> Installing Host orgrimmar Step: Restart; Details: Restarting vdsmd service. 
 <date>, <time> Installing Host orgrimmar Step: VDS Configuration. 
 <date>, <time> Installing Host orgrimmar Step: cleanAll. 
 <date>, <time> Installing Host orgrimmar Step: CoreDump. 
 <date>, <time> Installing Host orgrimmar Step: instCert. 
 <date>, <time> Installing Host orgrimmar Step: RHEV_INSTALL. 
 <date>, <time> Installing Host orgrimmar Step: Encryption setup. 
 <date>, <time> Installing Host orgrimmar Step: RHEV_INSTALL; Details: Connected to Host 192.168.1.1 with SSH key fingerprint: <the SSH fingerprint>. 
 <date>, <time> orgrimmar.azeroth.net parameters were updated by admin@internal. 
 <date>, <time> Power Management is not configured for Host orgrimmar 
 <date>, <time> Host orgrimmar.azeroth.net  registered. 
 <date>, <time> Host orgrimmar.azeroth.net  was added by UserName. 
 <date>, <time> Power Management is not configured for Host orgrimmar.azeroth.net
  • After the approval and registration some junk from the failed ssh key exchange issues will be left in the authorized_keys file of the :

Usually looks like this: 

<!DOCTYPE HTML PUBLIC "-//IETF//DTD HTML 2.0//EN">
<html><head>
<title>400 Bad Request</title>
</head><body>
<h1>Bad Request</h1>
<p>Your browser sent a request that this server could not understand.<br />
Reason: You're speaking plain HTTP to an SSL-enabled server port.<br />
Instead use the HTTPS scheme to access this URL, please.<br />
<blockquote>Hint: <a href="https://ovirtfoo.ctt.med.ge.com/"><b>https://ovirtfoo.ctt.med.ge.com/</b></a></blockquote></p>
<hr>
<address>Apache/2.2.22 (Fedora) Server at ovirtfoo.ctt.med.ge.com Port 443</address>
</body></html>
  • Delete the everything but the engine's SSH RSA key that you copied in previously in the hosts authorized_keys file
Retrieved from "http://www.ovirt.org/index.php?title=Using_EL6_hosts_with_Ovirt&oldid=4441"