cobbler and dnsmasq
by Gianluca Cecchi
Hello,
I have only one, but big server where I'm experimenting many things.
It is a Fedora 11 x86_64 host.
At first I configured it as a host for qemu/kvm guests, managing them with
the virt-manager gui tool.
I successfully had dnsmasq instance run by libvirt layer coexist with the
system provided dnsmasq daemon
This way I could manually PXE boot and install guests configured with shared
physical networks (bridge br0 on the host).
(I have two dnsmasq processes listening on different interfaces, br0 and
virbr0)
Now I'm looking at cobbler/koan to glue together things and provide also
provisioning for physical and virtual servers. Unfortunately I have to
configure/manage it on the same big box....
I configured cobbler to use dnsmasq and I thought that this way it would
pick up the already configured one.
Instead it seems to me that it overwrote completely my previous
/etc/dnsmasq.conf file ....
Is there any way to let cobbler use an externally configured dnsmasq or by
design it has to manage it?
At this moment I have
in /etc/cobbler/modules.conf:
[dhcp]
module = manage_dnsmasq
in /etc/cobbler/settings:
manage_dhcp: 1
Is it correct that I can preserve my already setup configs and settings only
putting manage_dhcp: 0?
this way cobbler will take advantage (or disadvantage if I scrumble things..
;-) and not interfere?
cobbler check gives:
[root@virtfed etc]# cobbler check
/usr/lib64/python2.6/site-packages/Cheetah/CacheRegion.py:30:
DeprecationWarning: the md5 module is deprecated; use hashlib instead
import md5
The following potential problems were detected:
#0: change 'disable' to 'no' in /etc/xinetd.d/tftp
#1: since iptables may be running, ensure 69, 80, and 25151 are unblocked
#2: The default password used by the sample templates for newly installed
machines (default_password_crypted in /etc/cobbler/settings) is still set to
'cobbler' and should be changed
#3: fencing tools were not found, and are required to use the (optional)
power management features. install cman to use them
In particular, I see that it configured this way
[root@virtfed etc]# cat /etc/dnsmasq.conf
# Cobbler generated configuration file for dnsmasq
# Wed Jul 1 10:02:47 2009
# resolve.conf .. ?
#no-poll
#enable-dbus
read-ethers
addn-hosts = /var/lib/cobbler/cobbler_hosts
dhcp-range=192.168.1.5,192.168.1.200
dhcp-option=3,10.4.5.152
dhcp-lease-max=1000
dhcp-authoritative
dhcp-boot=pxelinux.0
dhcp-boot=net:normalarch,pxelinux.0
dhcp-boot=net:ia64,/var/lib/cobbler/elilo-3.6-ia64.efi
Where does it pick the 192.168 address range....? Suppose I want to have
cobbler manage dhcp, where I configure the range? If I modify dnsmasq.conf
and then run "cobbler sync" do I risk to overwrite again my changes?
Eventually it could be nice to have cobbler making a backup copy (luckily I
had it) before overwrite....
I think that overwrite happened when I ran
cobbler import --path=/mnt --name=rhel4.8 --arch=x86
after mounting an iso of rhel4.8 x86....
btw: what is the difference between --path and --mirror? Perhaps the former
can only manage local paths, while the latter also Uri such as http/nfs...?
I didn't find clear answer...
I'm glad to say that anyway I was then able to install a guest on the same
host provinding cobbler with the command
koan --server=10.4.5.152 --virt --profile=rhel4.8-i386-virt
--virt-name=koantest
and then access it form within virt-manager. My compliments to you guys!
some annotations:
- even if in my kickstart I had
network --device=eth0 --bootproto static --ip 10.4.5.154 --netmask
255.255.255.0 --gateway 10.4.5.250 --nameserver 10.4.1.120 --hostname
koantest
It asked the network parameters in anaconda......
I presume because the kernel command line "automatically" contained
"ksdevice=link" but I'm not sure.
Also, this is a constraint with network installs and my line above is parsed
only to configure network for the system but not used at the initial
steps....
After putting the static network information, the installation proceeded
successfully.
- in /etc/cobbler settings I have
default_virt_type: qemu
but it seems cobbler configured the guest with the mac address used for xen
(initial 00:16:3e) instead of the one for qemu (52:54:00)
Thanks for your time,
Gianluca
14 years, 10 months
Whats the difference between cobbler.x86_64 and cobbler.noarch ?
by Anthony Giggins
Whats the difference between cobbler.x86_64 and cobbler.noarch ?
Name : cobbler
Arch : noarch
Version: 1.6.2
Release: 1.el5
Size : 938 k
Repo : epel
Summary: Boot server configurator
Description:
Cobbler is a network install server. Cobbler
supports PXE, virtualized installs, and
reinstalling existing Linux machines. The last two
modes use a helper tool, 'koan', that
integrates with cobbler. Cobbler's advanced features
include importing distributions from DVDs and rsync
mirrors, kickstart templating, integrated yum
mirroring, and built-in DHCP/DNS/power Management. Cobbler has
a Python and XMLRPC API for integration with other
applications. There is also a web interface.
Name : cobbler
Arch : x86_64
Version: 1.6.6
Release: 1.el5
Size : 921 k
Repo : epel
Summary: Boot server configurator
Description:
Cobbler is a network install server. Cobbler
supports PXE, virtualized installs, and
reinstalling existing Linux machines. The last two
modes use a helper tool, 'koan', that
integrates with cobbler. Cobbler's advanced features
include importing distributions from DVDs and rsync
mirrors, kickstart templating, integrated yum
mirroring, and built-in DHCP/DNS/power Management. Cobbler has
a Python and XMLRPC API for integration with other
applications. There is also a web interface.
Cheers,
Anthony
14 years, 10 months
cobbler buildiso and isolinux.bin
by Gianluca Cecchi
cobbler buildiso gives this error:
...
- using/creating tempdir: /root/buildiso
- building tree for isolinux
- copying miscellaneous files
Required file not found: /usr/lib/syslinux/isolinux.bin
My system is F11 x86_64 with syslinux-3.75-1.fc11.x86_64 and
cobbler-1.6.6-1.fc11.x86_64
My temporary workaround was to create a link like this
[root@virtfed ~]# ln -s /usr/share/syslinux /usr/lib/syslinux
and the command successfully completes and target machine is able to
boot from the kickstart.iso image (as a virtual media) and install
itself.
Probably the path cobbler has to query needs to be changed/extended?
Thanks,
Gianluca
14 years, 10 months
where can I get ipmitool for Centos 5.3?
by Anthony Giggins
where can I get ipmitool for Centos 5.3?
or is that even the right tool for DELL Poweredge R300 servers?
I can connect to the BMC via impish and issue commands successfully
Cheers,
Anthony
14 years, 10 months
variable help
by Tom Brown
I am trying to simplify adding some ETH_OPTS and basically what i need
is some way so that when an interface has say bond0 as its
bonding_master then add the following opts.
I can get that info with
#set eth0_opts = $getVar("bonding_master_eth0","")
#set eth1_opts = $getVar("bonding_master_eth1","")
#set eth2_opts = $getVar("bonding_master_eth2","")
#set eth3_opts = $getVar("bonding_master_eth3","")
#set eth4_opts = $getVar("bonding_master_eth4","")
#set eth5_opts = $getVar("bonding_master_eth5","")
#set eth6_opts = $getVar("bonding_master_eth6","")
#if eth0_opts == "bond0"
echo ETHTOOL_OPTS="speed 100 duplex full autoneg off" >>
/etc/sysconfig/network-scripts/ifcfg-eth0
etc etc
however where there are multiple interfaces that have bond0 as the
master this comes up short as it matches the first and then stops. How
can i make the 'if' carry on through all the other interfaces, or is
there a neater way to do this altogether ?
thanks
14 years, 10 months
Re: Adding a Distribution without copying the contents over?
by Robert Knight
>
> I see value in supporting the following:
> # cobbler distro add --kernel=http://path/to/vmlinuz
> --initrd=http://path/to/initrd.img --name=foo
>
> This command would copy vmlinuz and initrd.img in the correct places
> under /tftpboot making them available for PXE booting.
> It would allow people to use machines on the Internet (via http)
> instead of being forced to copy things locally (via NFS or other
> means).
I believe that you are asking for a change in the semantics of cobbler
distro add (or perhaps cobbler import). Changing command semantics
usually has unexpected results.
I've recently found that I want to do the same thing and that the
following works:
1. Copy the kernel and initrd to some place local with, say, wget.
2. Define the distribution with a command pointing to the local files
and add a ksmeta defining the tree at the http location.
Now, this is a (slight) abuse of the machinery, but will get the job
done without copying more than you need locally.
I hope this is helpful.
On Mon, 6 Jul 2009 16:00:41, Paul Company <pjcster(a)gmail.com> wrote:
> On the "how something works" front:
> I see value in supporting the following:
> # cobbler distro add --kernel=http://path/to/vmlinuz
> --initrd=http://path/to/initrd.img --name=foo
>
> This command would copy vmlinuz and initrd.img in the correct places
> under /tftpboot making them available for PXE booting.
> It would allow people to use machines on the Internet (via http)
> instead of being forced to copy things locally (via NFS or other
> means). Have you ever tried to NFS mount something from a remote
> machine on the Internet? Doesn't work.
>
> ON the "what to do" front:
> I currently have this issue at work, where our distros (vmlinuz &
> initrd.img) are available via http but they are not available via NFS
> (and will never be).
14 years, 10 months
Re: Adding a Distribution without copying the contents over?
by Simon Woolsgrove
>--- pjcster(a)gmail.com wrote:
>From: Paul Company <pjcster(a)gmail.com>
>To: cobbler mailing list <cobbler(a)lists.fedorahosted.org>
>Subject: Re: Adding a Distribution without copying the contents over?
>Date: Sat, 4 Jul 2009 23:05:47 -0700
>When you say "see them" you mean "ability to access them via a
>transport protocol" or what I liked to call an "access method". NFS is
>just one type of "access method". Wouldn't it be good to support http,
>https, ftp, rsync, ssh. Have you ever tried to NFS mount something on
>the Internet.
>If you see my original post, I thought "--available-as=nfs://etc."
>would do as you described, but it doesn't.
>So what is the difference between using --available-as=nfs:// and NFS
>mounting the path and then using import on the nfs mount point?
>Paul
I think this is the case
--available-as points to the location rather than importing the entire tree, which I think you want, vs the latter which would import the entire tree.
Cobbler creates a ks_meta value tree which is substitues (in the example ks for this path)
>From your post
# mkdir -p /distros/CentOS-5.3/x86_64
# cobbler import --path=/distros/CentOS-5.3/x86_64 --arch=x86_64
--name=CentOS-5.3-x86_64 \
--available-as=nfs://myserver1:/media
or
# cobbler import --path=/distros/CentOS-5.3/x86_64 --arch=x86_64
--name=CentOS-5.3-x86_64 \
--available-as=http://myserver1/media
Does /distros/CentOS-5.3/x86_64 have an actual copy of the media, as cobbler still needs to find the boot initrd, kernel, yum repo's etc ?
It should probably give a message that it does not find anything...
As others have pointed you do not have to use import and could use distro/profile and use your own ks_meta variables to locate the data tree whci can be compeltely seperate from the cobbler server if thats what you want.
Cheers,
Simon
14 years, 10 months
Learn to Hack on Cobbler week begins!
by Michael DeHaan
I hope everyone is exited about Learn to Hack on Cobbler week, which is
this week, all week, July 6th-10th. All questions are fair game, and
I'm going to be making myself fully available to helping you learn to
add features to Cobbler and answer all sorts of development questions.
If you can access IRC, you may want to join #cobbler-devel on
irc.freenode.net.
If you can't, Q&A here is absolutely fine too, and we'll try to respond
very quickly ... though you might want to try something like
http://www.mibbit.com/chat/ (haven't used it, looks like it may be
helpful for works that block IRC though) first.
--Michael
14 years, 10 months
Custom Kickstart not completely honoured with Centos 5.3
by Anthony Giggins
Hi Everyone,
We use a custom kickstart to roll out our linux SOE, for some reason
some of the the config doesn't get honoured
(anything below $SNIPPET('redhat_register')) with Centos 5.3 Installs
but is does with 5.2 can anyone advise what is wrong?
Below is a copy of the kickstart config that's in cobbler with any
passwords etc removed, was originally in the message but it was too big
for the mailing list.
http://pastebin.com/f3e5a1909
Cheers,
Anthony
14 years, 10 months
Adding a Distribution without copying the contents over?
by Paul Company
On "myserver1" I mounted CentOS-5.3-x86_64-bin-DVD.iso on /media and
NFS exported /media.
I also make /media available via Apache - http://myserver1/media/
Then I ran the following on "myserver2", which has cobbler installed.
# mkdir -p /distros/CentOS-5.3/x86_64
# cobbler import --path=/distros/CentOS-5.3/x86_64 --arch=x86_64
--name=CentOS-5.3-x86_64 \
--available-as=nfs://myserver1:/media
or
# cobbler import --path=/distros/CentOS-5.3/x86_64 --arch=x86_64
--name=CentOS-5.3-x86_64 \
--available-as=http://myserver1/media
I get the following output:
---------------- (adding distros)
---------------- (associating kickstarts)
But, when I run the following:
# cobbler list
There's no output!!!!
Isn't this supposed to make the distro available for use?
I'd like to add the distro (CentOS-5.3-x86_64-bin-DVD.iso) but I don't
want to copy the contents over.
Seems like "import" only works if things get copied locally?
Can cobbler work without having the distro locally on the cobbler server?
Paul
14 years, 10 months