4:12 a.m.
--------------050608080707060402090909
Content-Type: text/plain; charset="UTF-8"; format=flowed
Content-Transfer-Encoding: 7bit
On 12/13/2013 2:59 AM, Sven Kieske wrote:
Hi,
first, thanks for this script!
I'll have to setup some CentOS 6.5 machines too, maybe it will help.
Here are some questions/improvements from me:
There's no need to use "localinstall" anymore, "install" is fine
with
yum :-)
Then you install "virt-manager", for what purpose, may I ask?
I also never needed to manually create the ovirt-management bridge
has this behaviour changed in recent vdsm/CentOS Release?
Am 13.12.2013 05:48, schrieb Ted Miller:
> # script to prepare Centos 6.5 for ovirt host install process
>
> echo "================= Ted's personal preferences--early
========================"
>
> yum -y install nano deltarpm yum-plugin-priorities yum-presto mlocate
>
> echo "=================== end of Ted's personal preferences--early
================"
>
> yum -y upgrade
>
> echo "....................install some repos (if not already
done).................."
> cd /etc/yum.repos.d
> if [ ! -f glusterfs-epel.repo ] ; then
> echo "..............installing gluster repo......................."
> yum -y install wget
> wget
http://download.gluster.org/pub/gluster/glusterfs/LATEST/EPEL.repo/gluste...
> echo "..............done installing gluster repo.................."
> fi
>
> if [ ! -f el6-ovirt.repo ] ; then
> echo "..............installing ovirt repo......................."
> yum -y localinstall http://ovirt.org/releases/ovirt-release-el.noarch.rpm
> echo "..............done installing ovirt repo.................."
> fi
>
> if [ ! -f epel.repo ] ; then
> echo "..............installing epel repo......................."
> yum -y localinstall
http://mirror.us.leaseweb.net/epel/6/i386/epel-release-6-8.noarch.rpm
> echo "..............done installing epel repo.................."
> fi
>
>
> echo "........install libvirt............................"
> yum -y install libvirt qemu-kvm tuned
> echo "........install virt-manager" # with unlisted dependencies
> yum -y install virt-manager xorg-x11-xauth dejavu-lgc-sans-mono-fonts
>
> # create the ovirtmgmt bridge
> if [ ! -f /etc/sysconfig/network-scripts/ifcfg-ovirtmgmt ]; then
> echo "........................creating ovirtmgmt
bridge......................"
> service libvirtd start
> service libvirtd status
> virsh net-destroy default
> virsh net-undefine default
> virsh iface-bridge eth0 ovirtmgmt
> service network restart
> service libvirtd stop
> service libvirtd status
> fi
>
> echo ".........copy tuned profile..............."
> # copy virtual-host --> rhs-virtualization so ovirt is happy
> cp -r /etc/tune-profiles/virtual-host /etc/tune-profiles/rhs-virtualization
>
> yum -y install vdsm
>
> echo "================= Ted's personal preferences--late
================================="
>
> #add lines to send messages to TTY12
> cat /etc/rsyslog.conf | grep tty12
> if [ ! $? -eq 0 ] ; then
> echo ".......................Adding for tty12...................."
> echo " " >> /etc/rsyslog.conf
> echo "# Log everything to tty12" >> /etc/rsyslog.conf
> echo "*.* /dev/tty12" >>
/etc/rsyslog.conf
> service rsyslog restart
> fi
>
> echo "...........install gkrellm.............."
> yum -y install gkrellm
>
> #add poll=0 to kill noveau messages
> cat /boot/grub/grub.conf | grep poll=0
> if [ ! $? -eq 0 ] ; then
> echo "............................Adding
poll=0.........................."
> sed -i '/^[ \t]kernel*/ s/$/ drm-kms-helper.poll=0/g'
/boot/grub/grub.conf
> fi
>
I followed your instructions (and similar suggestions by others). I was
installing the host only, so commented out everything in my file except
install EPEL and ovirt repos. I'm not interested in hearing about what
"should" happen, or what happened on 6.4. These are the (unfixed) problems
with 6.5, and they are real.
INSTALL FAILED with message:
Failed to install Host office4a. Yum [u'glusterfs-server-3.4.0-8.el6.x86_64
requires glusterfs-libs = 3.4.0-8.el6', u'glusterfs-server-3.4.0-8.el6.x86_64
requires glusterfs = 3.4.0-8.el6', u'glusterfs-server-3.4.0-8.el6.x86_64
requires glusterfs-fuse = 3.4.0-8.el6', u'glusterfs-cli-3.4.0-8.el6.x86_64
requires glusterfs-libs = 3.4.0-8.el6'].
# yum list gluster
base: mirror.oss.ou.edu
epel: mirrors.servercentral.net
extras: mirror.dattobackup.com
updates: centos.sonn.com
Available Packages
glusterfs.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-api.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-api-devel.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-cli.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-debuginfo.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-devel.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-fuse.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-geo-replication.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-libs.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-rdma.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-resource-agents.noarch 3.4.0-8.el6 glusterfs-noarch-epel
glusterfs-server.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-ufo.noarch 3.4.0-8.el6 glusterfs-noarch-epel
The real truth is that there are three repos trying to serve up gluster
3.4.0 packages.
# yum list gluster* --showduplicates
Loaded plugins: fastestmirror, presto
Loading mirror speeds from cached hostfile
* base: mirror.umd.edu
* epel: mirror.utexas.edu
* extras: mirrors.liquidweb.com
* updates: centos.sonn.com
Available Packages
glusterfs.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs.x86_64 3.4.0-8.el6 ovirt-stable
glusterfs.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-api.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-api.x86_64 3.4.0-8.el6 ovirt-stable
glusterfs-api.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-api-devel.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-api-devel.x86_64 3.4.0-8.el6 ovirt-stable
glusterfs-api-devel.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-cli.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-cli.x86_64 3.4.0-8.el6 ovirt-stable
glusterfs-debuginfo.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-debuginfo.x86_64 3.4.0-8.el6 ovirt-stable
glusterfs-devel.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-devel.x86_64 3.4.0-8.el6 ovirt-stable
glusterfs-devel.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-fuse.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-fuse.x86_64 3.4.0-8.el6 ovirt-stable
glusterfs-fuse.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-geo-replication.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-geo-replication.x86_64 3.4.0-8.el6 ovirt-stable
glusterfs-libs.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-libs.x86_64 3.4.0-8.el6 ovirt-stable
glusterfs-libs.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-rdma.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-rdma.x86_64 3.4.0-8.el6 ovirt-stable
glusterfs-rdma.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-resource-agents.noarch 3.4.0-8.el6 glusterfs-noarch-epel
glusterfs-resource-agents.noarch 3.4.0-8.el6 ovirt-stable
glusterfs-server.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-server.x86_64 3.4.0-8.el6 ovirt-stable
glusterfs-ufo.noarch 3.4.0-8.el6 glusterfs-noarch-epel
I had found a suggestion on the web to upgrade to 3.4.1, since the 3.4.0
versions can't figure out "who's on first".
So I went back and uncommented the glusterfs repo install. That changes
yum's perspective to:
# yum list gluster*
Loaded plugins: fastestmirror, presto
Loading mirror speeds from cached hostfile
* base: mirrors.einstein.yu.edu
* epel: mirror.utexas.edu
* extras: mirror.nexcess.net
* updates: centos.sonn.com
Available Packages
glusterfs.x86_64 3.4.1-3.el6 glusterfs-epel
glusterfs-api.x86_64 3.4.1-3.el6 glusterfs-epel
glusterfs-api-devel.x86_64 3.4.1-3.el6 glusterfs-epel
glusterfs-cli.x86_64 3.4.1-3.el6 glusterfs-epel
glusterfs-debuginfo.x86_64 3.4.1-3.el6 glusterfs-epel
glusterfs-devel.x86_64 3.4.1-3.el6 glusterfs-epel
glusterfs-fuse.x86_64 3.4.1-3.el6 glusterfs-epel
glusterfs-geo-replication.x86_64 3.4.1-3.el6 glusterfs-epel
glusterfs-libs.x86_64 3.4.1-3.el6 glusterfs-epel
glusterfs-rdma.x86_64 3.4.1-3.el6 glusterfs-epel
glusterfs-resource-agents.noarch 3.4.1-3.el6 glusterfs-noarch-epel
glusterfs-server.x86_64 3.4.1-3.el6 glusterfs-epel
Only by replacing the glusterfs-epel repo file that ovirt uses with a
different one, does version 3.4.1 become available, and the error goes away.
I got a message saying "Host office4a installation in progress. Cannot set
tuned profile." but that message did not stop the installation from
completing, but gives me enough reason to put back the line copying the
"tuned" profile, so ovirt finds what it is looking for.
The other problems did not appear this time, very possibly because of the
order in which I tackled things, based on the advice I got.
That leaves my install script looking like this (with all the 'if' statements
take out):
#!/bin/bash
# script to prepare Centos 6.5 minimal for ovirt host install process
# by Ted Miller
# 2013-12-16 tested on office4a
yum -y install yum-presto
yum -y upgrade
yum -y install wget
wget
http://download.gluster.org/pub/gluster/glusterfs/LATEST/EPEL.repo/gluste...
yum -y install http://ovirt.org/releases/ovirt-release-el.noarch.rpm
yum -y install http://mirror.us.leaseweb.net/epel/6/i386/epel-release-6-8.noarch.rpm
cp -r /etc/tune-profiles/virtual-host /etc/tune-profiles/rhs-virtualization
That is significantly reduced from what I had, but still a long ways from the
"just install the epel and ovirt repos" that I was getting.
The version conflict that forced the addition of the glusterfs repo is
probably due to new versions of gluster in Centos 6.5.
The problem with "tune" was trying to fix a problem for RHES users, but
created a problem for everyone else (CentOS & etc).
Ted Miller
--------------050608080707060402090909
Content-Type: text/html; charset="UTF-8"
Content-Transfer-Encoding: 8bit
<html>
<head>
<meta content="text/html; charset=UTF-8"
http-equiv="Content-Type">
</head>
<body text="#000000" bgcolor="#FFFFFF">
On 12/13/2013 2:59 AM, Sven Kieske wrote:<br>
<blockquote cite="mid:52AABEF4.8000608@mittwald.de"
type="cite">
<pre wrap="">Hi,
first, thanks for this script!
I'll have to setup some CentOS 6.5 machines too, maybe it will help.
Here are some questions/improvements from me:
There's no need to use "localinstall" anymore, "install" is fine
with
yum :-)
Then you install "virt-manager", for what purpose, may I ask?
I also never needed to manually create the ovirt-management bridge
has this behaviour changed in recent vdsm/CentOS Release?
Am 13.12.2013 05:48, schrieb Ted Miller:
</pre>
<blockquote type="cite">
<pre wrap=""># script to prepare Centos 6.5 for ovirt host install
process
echo "================= Ted's personal preferences--early
========================"
yum -y install nano deltarpm yum-plugin-priorities yum-presto mlocate
echo "=================== end of Ted's personal preferences--early
================"
yum -y upgrade
echo "....................install some repos (if not already
done).................."
cd /etc/yum.repos.d
if [ ! -f glusterfs-epel.repo ] ; then
echo "..............installing gluster repo......................."
yum -y install wget
wget <a class="moz-txt-link-freetext"
href="http://download.gluster.org/pub/gluster/glusterfs/LATEST/EPEL....
echo "..............done installing gluster repo.................."
fi
if [ ! -f el6-ovirt.repo ] ; then
echo "..............installing ovirt repo......................."
yum -y localinstall <a class="moz-txt-link-freetext"
href="http://ovirt.org/releases/ovirt-release-el.noarch.rpm">...
echo "..............done installing ovirt repo.................."
fi
if [ ! -f epel.repo ] ; then
echo "..............installing epel repo......................."
yum -y localinstall <a class="moz-txt-link-freetext"
href="http://mirror.us.leaseweb.net/epel/6/i386/epel-release-6-8.noa...
echo "..............done installing epel repo.................."
fi
echo "........install libvirt............................"
yum -y install libvirt qemu-kvm tuned
echo "........install virt-manager" # with unlisted dependencies
yum -y install virt-manager xorg-x11-xauth dejavu-lgc-sans-mono-fonts
# create the ovirtmgmt bridge
if [ ! -f /etc/sysconfig/network-scripts/ifcfg-ovirtmgmt ]; then
echo "........................creating ovirtmgmt
bridge......................"
service libvirtd start
service libvirtd status
virsh net-destroy default
virsh net-undefine default
virsh iface-bridge eth0 ovirtmgmt
service network restart
service libvirtd stop
service libvirtd status
fi
echo ".........copy tuned profile..............."
# copy virtual-host --> rhs-virtualization so ovirt is happy
cp -r /etc/tune-profiles/virtual-host /etc/tune-profiles/rhs-virtualization
yum -y install vdsm
echo "================= Ted's personal preferences--late
================================="
#add lines to send messages to TTY12
cat /etc/rsyslog.conf | grep tty12
if [ ! $? -eq 0 ] ; then
echo ".......................Adding for tty12...................."
echo " " >> /etc/rsyslog.conf
echo "# Log everything to tty12" >> /etc/rsyslog.conf
echo "*.* /dev/tty12" >>
/etc/rsyslog.conf
service rsyslog restart
fi
echo "...........install gkrellm.............."
yum -y install gkrellm
#add poll=0 to kill noveau messages
cat /boot/grub/grub.conf | grep poll=0
if [ ! $? -eq 0 ] ; then
echo "............................Adding poll=0.........................."
sed -i '/^[ \t]kernel*/ s/$/ drm-kms-helper.poll=0/g' /boot/grub/grub.conf
fi
</pre>
</blockquote>
</blockquote>
<div class="moz-text-html" lang="x-western">
<div name="divtagdefaultwrapper" id="divtagdefaultwrapper"
style="font-family: Calibri,Arial,Helvetica,sans-serif;
font-size: 12pt; color: #000000; margin: 0"> I followed your
instructions (and similar suggestions by others). I was
installing the host only, so commented out everything in my file
except install EPEL and ovirt repos. I'm not interested in
hearing about what "should" happen, or what happened on 6.4.
These are the (unfixed) problems with 6.5, and they are real.<br>
<br>
INSTALL FAILED with message:<br>
<br>
Failed to install Host office4a. Yum
[u'glusterfs-server-3.4.0-8.el6.x86_64 requires glusterfs-libs =
3.4.0-8.el6', u'glusterfs-server-3.4.0-8.el6.x86_64 requires
glusterfs = 3.4.0-8.el6', u'glusterfs-server-3.4.0-8.el6.x86_64
requires glusterfs-fuse = 3.4.0-8.el6',
u'glusterfs-cli-3.4.0-8.el6.x86_64 requires glusterfs-libs =
3.4.0-8.el6'].<br>
<br>
<pre><tt># yum list gluster
base: mirror.oss.ou.edu
epel: mirrors.servercentral.net
extras: mirror.dattobackup.com
updates: centos.sonn.com
Available Packages
glusterfs.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-api.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-api-devel.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-cli.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-debuginfo.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-devel.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-fuse.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-geo-replication.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-libs.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-rdma.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-resource-agents.noarch 3.4.0-8.el6 glusterfs-noarch-epel
glusterfs-server.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-ufo.noarch 3.4.0-8.el6 glusterfs-noarch-epel</tt>
</pre>
<br>
The real truth is that there are three repos trying to serve up
gluster 3.4.0 packages.<br>
<br>
<pre><tt># yum list gluster* --showduplicates
Loaded plugins: fastestmirror, presto
Loading mirror speeds from cached hostfile
* base: mirror.umd.edu
* epel: mirror.utexas.edu
* extras: mirrors.liquidweb.com
* updates: centos.sonn.com
Available Packages
glusterfs.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs.x86_64 3.4.0-8.el6 ovirt-stable
glusterfs.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-api.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-api.x86_64 3.4.0-8.el6 ovirt-stable
glusterfs-api.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-api-devel.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-api-devel.x86_64 3.4.0-8.el6 ovirt-stable
glusterfs-api-devel.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-cli.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-cli.x86_64 3.4.0-8.el6 ovirt-stable
glusterfs-debuginfo.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-debuginfo.x86_64 3.4.0-8.el6 ovirt-stable
glusterfs-devel.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-devel.x86_64 3.4.0-8.el6 ovirt-stable
glusterfs-devel.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-fuse.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-fuse.x86_64 3.4.0-8.el6 ovirt-stable
glusterfs-fuse.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-geo-replication.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-geo-replication.x86_64 3.4.0-8.el6 ovirt-stable
glusterfs-libs.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-libs.x86_64 3.4.0-8.el6 ovirt-stable
glusterfs-libs.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-rdma.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-rdma.x86_64 3.4.0-8.el6 ovirt-stable
glusterfs-rdma.x86_64 3.4.0.36rhs-1.el6 base
glusterfs-resource-agents.noarch 3.4.0-8.el6 glusterfs-noarch-epel
glusterfs-resource-agents.noarch 3.4.0-8.el6 ovirt-stable
glusterfs-server.x86_64 3.4.0-8.el6 glusterfs-epel
glusterfs-server.x86_64 3.4.0-8.el6 ovirt-stable
glusterfs-ufo.noarch 3.4.0-8.el6
glusterfs-noarch-epel</tt><font face="'Lucida Console', Monaco,
monospace">
</font></pre>
<br>
<br>
I had found a suggestion on the web to upgrade to 3.4.1, since
the 3.4.0 versions can't figure out "who's on first".<br>
<br>
So I went back and uncommented the glusterfs repo install. That
changes yum's perspective to:<br>
<br>
<pre><tt># yum list gluster*
Loaded plugins: fastestmirror, presto
Loading mirror speeds from cached hostfile
* base: mirrors.einstein.yu.edu
* epel: mirror.utexas.edu
* extras: mirror.nexcess.net
* updates: centos.sonn.com
Available Packages
glusterfs.x86_64 3.4.1-3.el6 glusterfs-epel
glusterfs-api.x86_64 3.4.1-3.el6 glusterfs-epel
glusterfs-api-devel.x86_64 3.4.1-3.el6 glusterfs-epel
glusterfs-cli.x86_64 3.4.1-3.el6 glusterfs-epel
glusterfs-debuginfo.x86_64 3.4.1-3.el6 glusterfs-epel
glusterfs-devel.x86_64 3.4.1-3.el6 glusterfs-epel
glusterfs-fuse.x86_64 3.4.1-3.el6 glusterfs-epel
glusterfs-geo-replication.x86_64 3.4.1-3.el6 glusterfs-epel
glusterfs-libs.x86_64 3.4.1-3.el6 glusterfs-epel
glusterfs-rdma.x86_64 3.4.1-3.el6 glusterfs-epel
glusterfs-resource-agents.noarch 3.4.1-3.el6 glusterfs-noarch-epel
glusterfs-server.x86_64 3.4.1-3.el6 glusterfs-epel
</tt></pre>
<tt> </tt><br>
Only by replacing the glusterfs-epel repo file that ovirt uses
with a different one, does version 3.4.1 become available, and
the error goes away.<br>
<br>
I got a message saying "Host office4a installation in progress.
Cannot set tuned profile." but that message did not stop the
installation from completing, but gives me enough reason to put
back the line copying the "tuned" profile, so ovirt finds what
it is looking for.<br>
<br>
The other problems did not appear this time, very possibly
because of the order in which I tackled things, based on the
advice I got.<br>
<br>
That leaves my install script looking like this (with all the
'if' statements take out):<br>
<br>
<pre><font face="'Lucida Console', Monaco,
monospace"> </font><tt>#!/bin/bash
# script to prepare Centos 6.5 minimal for ovirt host install process
# by Ted Miller
# 2013-12-16 tested on office4a
yum -y install yum-presto
yum -y upgrade
yum -y install wget
wget <a class="moz-txt-link-freetext"
href="http://download.gluster.org/pub/gluster/glusterfs/LATEST/EPEL....
yum -y install <a class="moz-txt-link-freetext"
href="http://ovirt.org/releases/ovirt-release-el.noarch.rpm">...
yum -y install <a class="moz-txt-link-freetext"
href="http://mirror.us.leaseweb.net/epel/6/i386/epel-release-6-8.noa...
cp -r /etc/tune-profiles/virtual-host
/etc/tune-profiles/rhs-virtualization</tt><font face="'Lucida
Console', Monaco, monospace">
</font></pre>
<br>
That is significantly reduced from what I had, but still a long
ways from the "just install the epel and ovirt repos" that I was
getting.<br>
<br>
The version conflict that forced the addition of the glusterfs
repo is probably due to new versions of gluster in Centos 6.5.<br>
The problem with "tune" was trying to fix a problem for RHES
users, but created a problem for everyone else (CentOS &
etc).<br>
<br>
Ted Miller<br>
<br>
</div>
</div>
</body>
</html>
--------------050608080707060402090909--