PXE Boot - blacklist=
by Alex Davies
Hi All,
I am trying to kickstart two sun "Thumper" boxes. These boxes have QLogic
ISCSI cards in them, which take over an hour to initialize during kickstart.
This is very tedious.
Since I am not installing to a ISCSI volume, I would like to blacklist this
driver using the --kopts=" blacklist=qla4xxx "
[root@gcisvlin-cobbler ~]# cd /tftpboot/pxelinux.cfg/
[root@gcisvlin-cobbler pxelinux.cfg]# grep -irn blacklist *
01-00-14-4f-f2-9e-b0:6: append initrd=/images/rhel5-x86_64/initrd.img
ksdevice=eth0 lang= text blacklist=qla4xxx syslog=10.5.10.5:25150kssendmac ks=
http://10.5.10.5/cblr/svc/op/ks/system/gcisvxen011xa
01-00-14-4f-f2-b7-68:6: append initrd=/images/rhel5-x86_64/initrd.img
ksdevice=eth0 lang= text blacklist=qla4xxx syslog=10.5.10.5:25150kssendmac ks=
http://10.5.10.5/cblr/svc/op/ks/system/gcisvxen011xb
However, this does not work - the driver still loads during kickstart.
While this isnt the end of the world, i'd love to know if there is something
dumb that I am doing.
Many thanks,
Alex
15 years, 2 months
Re: network ordering
by Andreas Zuber
Am Montag 16 März 2009 20:25:21 schrieb Jasper Capel:
> > bootif and ipappend 2 is already the default in newer cobbler installs,
> > I'm thinking that section of the Wiki no longer needs to be there?
> >
> > --Michael
>
> I've just removed that entire section. Some things are already the
> defaults, and the rest is outdated (the snippet, the functionality was
> adopted though, in what is now post_install_network_config).
Ok, seams like I'm also outdated :-/ thx for the update :-)
--
Andreas Zuber
Linux System-Ingenieur
Puzzle ITC GmbH
www.puzzle.ch
Telefon +41 31 370 22 00
Direkt +41 31 370 22 49
Mobile +41 79 766 25 51
Fax +41 31 370 22 01
Puzzle ist Mitglied der ODF Alliance:
<http://www.puzzle.ch/odfalliance/>
15 years, 2 months
network ordering
by Christopher Johnston
So I have some servers with a few network cards installed and because the
kernel does a poor job of ordering them correctly during an install they
come up out of order (pci=bfsort does not help). I havnt looked to deep
into the snippet just yet, but on a specific system where I have the
following:
eth0 : unconfigured
eth1 : unconfigured
eth2 : non-routeable network (heartbeat)
eth3 : configure routeable network
My BIOS is configured to PXE boot on eth3. So in cobbler I have it setup
like so with eth0 empty, eth2 configured our private 192.168 interface and
eth3 configured with the routable interface.
interface : eth0
mac address :
bonding :
bonding_master :
bonding_opts :
is static? : False
ip address :
subnet :
static routes : []
dns name :
dhcp tag :
virt bridge :
interface : eth2
mac address : 00:1F:29:E1:54:AE
bonding :
bonding_master :
bonding_opts :
is static? : True
ip address : 192.168.x.x
subnet : 255.255.255.0
static routes : []
dns name : edge01.core.lon04.mtt
dhcp tag :
virt bridge :
interface : eth3
mac address : 00:1F:29:E1:54:AC
bonding :
bonding_master :
bonding_opts :
is static? : True
ip address : 10.x.x.x
subnet : 255.255.255.0
static routes : []
dns name : edge01.lon04.mtt
dhcp tag :
virt bridge :
When I look at the kickstart file, it looks to be ordering them with a
different configuration then I wanted using the "old" style config? how do
I get it to use the "new" style config?
# Network information
# Using "old" style networking config. Make sure all MAC-addresses are in
cobbler to use the new-style config
network --bootproto=static --ip=10.123.200.36 --netmask=255.255.255.0
--gateway=10.123.200.1 --hostname=edge01.lon04.mtt --device=eth0 --onboot=on
network --bootproto=static --ip=192.168.123.36 --netmask=255.255.255.0
--gateway=10.123.200.1 --hostname=edge01.lon04.mtt --device=eth1 --onboot=on
network --bootproto=dhcp --device=eth2 --onboot=on
A few questions:
1) eth0 should not even be used, let alone why is it a requirement to have
eth2 in my cobbler system record? I have alot of hosts where eth0 is
available but not even plumbed up.
2) eth2/eth3 are the NICs I want to use, but why did they end up going to
eth0/eth1?
3) where did the eth2 interface come into the picture here using dhcp?
15 years, 2 months
Xen?
by drew einhorn
Do we know where Red Hat is going with virtualization?
Recent announcements indicate that Xen is being abandoned.
Do we know what will replace Xen?
And what it will take for cobbler to support it?
--
Drew Einhorn
15 years, 2 months
Mac ranges
by Michael DeHaan
Cobbler currently generates random MAC ranges for Xen in the assigned
XenSource space.
The problem is, from what I can tell, the qemu space (which we should be
using is much smaller...
macaddr[0] = 0x52;
macaddr[1] = 0x54;
macaddr[2] = 0x00;
macaddr[3] = 0x12;
macaddr[4] = 0x34;
macaddr[5] = 0x56 + idx;
This doesn't look very good for nice enterprisey KVM usage when we're
managing DHCP and DNS and have hundreds of system records.
Should we just continue using the Xen range? Or is there a "private"
space we should be using?
VMware requires it's own space, so what little VMware support we have in
koan should be generating macs in those ranges.
The idea is to make one XMLRPC function (we have one now) that instead
of generating a random mac in the Xen range that does not yet conflict,
generates one in the appropriate range that does not yet conflict.
--Michael
15 years, 2 months
baffling problem with cobbler kickstart
by Terry McIntyre
When I check the cobbler-generated kickstart via "cobbler system getks
--name=sapcps", I get this value for the url line:
url --url=http://192.168.63.57:80/cblr/links/Redhat4x64
When the same kickstart is summoned via the kickstart process, the url line is
url --url=$tree
I can't figure out why the file differs depending on the invocation.
--
Terry McIntyre
UNIX for hire
Software Development, Systems Administration, Security
terry.mcintyre(a)gmail.com
15 years, 2 months
git integration and other ways to version control config changes
by Michael DeHaan
Yesterday Jeff S and I were talking on IRC about making a trigger that
would be called on any edit command, which could fire off version
control commands, such as git.
I think I'm going to do this in settings:
versioned_config_enable: 0/1
versioning_config_options: { "engine" : "git" }
Initially git will be the only valid choice, though folks can patch the
"generic edit" trigger I'm going to make to add other types and options,
if needed.
The question comes up as to whether we version control just
"/var/lib/cobbler/config" or we also try to do
"/var/lib/cobbler/kickstarts". However there's really not a good
interval to do VC
on kickstarts, but we could do it any time things were edited and/or
synced, pretty easily. This would mean we'd need to ship (and
install) a .gitignore in /var/lib/cobbler just to make sure
we didn't try to version track any of the bootloaders.
Once we do this we open up a slew of nice features, including easy
rollbacks when someone makes a change they shouldn't have made, and RSS
feeds of changes and such.
So, any thoughts on how much to track or the above ideas?
I'm going to start small, so we could extend this later, I'm sure.
Though I'd want to start out version controlling the right root dir, and
I'm not sure if that should be /var/lib/cobbler/config or all of
/var/lib/cobbler.
If we do all of /var/lib/cobbler, we'll easily be able to track lots of
good changes, the only thing to note is that the snapshot times for the
kickstarts and human edited (not API edited files) may not have
timestamps that correlate with when they were edited. I think that's ok.
I don't particularly want to introduce an inotify (a kernel system of
being able to be notified when things change) process just for that
action, but ultimately, this /is/ something we could do.
--Michael
15 years, 2 months
Re: Using koan --replace-self problem w/ F10 and 11 - but not 9
by Scot Floess
OK, so I have my machine up and running right now...
All the dialog box states is "There is an error configuring your network interface." And there is a retry button...
alt-f3 shows me things like the kernel command used (for example a param is ksdevice=eth0), "text mode forced from cmdline", "early networking required for syslog", etc. There is a fatal error regarding the floppy - "No such device" - which, of course there is not... I also see "getting kickstart file" and "doing kickstart... seting it up"
alt-f4 shows alot of stuff...scrolled by so didnt catch it all... But I do see "NetworkManager bringing up device..."line up, 100Mbps, full-duplex, lap 0x45E1" "deactivating device (reason: 2)", "carrier now on (device state 2)", "device state change: 2-> 3"
While all this is transpiring, alt-f1 has a dialog box stating "Waiting for NetworkManager to configure eth0"
Then, ultimately I get the "There is an error configuring your network interface." on the alt-f1 screen...
When I click the retry button, I see ifcg-fedora statements that look successful in alt-f4...
And it starts the install..
Its not really a big deal...just a minor annoyance... Heck I'm psyched to be able to install F10 after so many weeks of messing around :)
----- Original Message -----
From: "Scot Floess" <sfloess(a)redhat.com>
To: "cobbler mailing list" <cobbler(a)lists.fedorahosted.org>
Sent: Thursday, March 12, 2009 9:41:02 AM GMT -05:00 US/Canada Eastern
Subject: Re: Using koan --replace-self problem w/ F10 and 11 - but not 9
So, to be honest, I was at work till 11pm last night...got home around midnight and was anxious to try this out :) Being I was somewhat bleary eyed...I don't remember the exact error as it was stated in the dialog box. If memory serves, I saw some mention of the NetworkManager doing some type of work and then the dialog box proclaiming something about a network error - and there being a retry button - which I depressed...
I can definitely try it this evening (or tomorrow) and write down what I see.
BTW, this wasn't for a VM - I was doing a bare metal "koan --replace-self" installation...
----- Original Message -----
From: "Michael DeHaan" <mdehaan(a)redhat.com>
To: "cobbler mailing list" <cobbler(a)lists.fedorahosted.org>
Sent: Thursday, March 12, 2009 9:35:35 AM GMT -05:00 US/Canada Eastern
Subject: Re: Using koan --replace-self problem w/ F10 and 11 - but not 9
Scot Floess wrote:
> I just wanted to let y'all know...
>
> Changing the kernel option to /var/www/cobbler/ks_mirror/CentOS-5.2-i386/images/pxeboot/vmlinuz did, in fact, fix my problem.
>
> Fedora initially complained with a retry dialog box...but selecting that...started the install as I expected...
>
> So its working for me...
>
> I really, want to thank everyone for helping me out! I would have never figured it out...
>
What did the retry box say?
(I've seen some issues with F10 installs in virt guests and network
manager, though newer versions of koan send the OS info to libvirt so
the right virtual interface can be chosen.... haven't seen any retry
boxes with regular physical installs.)
> ----- Original Message -----
> From: "Scot Floess" <sfloess(a)redhat.com>
> To: "cobbler mailing list" <cobbler(a)lists.fedorahosted.org>
> Sent: Wednesday, March 11, 2009 12:35:20 PM GMT -05:00 US/Canada Eastern
> Subject: Re: Using koan --replace-self problem w/ F10 and 11 - but not 9
>
> Oh wow... I get it :)
>
> Yeah - my home network is full of top-of-the-line machines - circa 2000 ;) Hee - none of them support PAE :)
>
> Again, thanks y'all for all your help!
>
> ----- Original Message -----
> From: "Michael DeHaan" <mdehaan(a)redhat.com>
> To: "cobbler mailing list" <cobbler(a)lists.fedorahosted.org>
> Sent: Wednesday, March 11, 2009 12:18:32 PM GMT -05:00 US/Canada Eastern
> Subject: Re: Using koan --replace-self problem w/ F10 and 11 - but not 9
>
> Scot Floess wrote:
>
>> So, for my CentOS 5.2, I see the following entries:
>>
>> Kernel: /var/www/cobbler/ks_mirror/CentOS-5.2-i386/images/pxeboot/vmlinuz
>> Initrd:
>> /var/www/cobbler/ks_mirror/CentOS-5.2-i386/images/pxeboot/initrd.img
>>
>> For Fedora 10, I see:
>>
>> Kernel:
>> /var/www/cobbler/ks_mirror/Fedora-10-i386/images/pxeboot/vmlinuz-PAE
>> Initrd:
>> /var/www/cobbler/ks_mirror/Fedora-10-i386/images/pxeboot/initrd.img
>>
>> I also see that the file
>> /var/www/cobbler/ks_mirror/Fedora-10-i386/images/pxeboot/vmlinuz exists...
>>
>> So do I need to change Kernel for Fedora 10 to be
>> var/www/cobbler/ks_mirror/Fedora-10-i386/images/pxeboot/vmlinuz ???
>>
>> I'd try it - but am not at home where my machines are :) Oh heck,
>> I'll definitely try it tonight for sure :D
>>
>>
>>
>
> Yes you can cobbler distro edit to fix that and you should be good to go.
>
> The import code is, I think, just globbing against "vmlinuz*" so it's
> getting the first one it finds, when it probably should create two
> distros and not one in this instance.
>
> I'm guessing your machine can't do PAE and that's why you have the problem.
>
>
> --Michael
>
>> ----- Original Message -----
>> From: "Ronald J. Yacketta" <yacketrj(a)potsdam.edu>
>> To: "cobbler mailing list" <cobbler(a)lists.fedorahosted.org>
>> Sent: Wednesday, March 11, 2009 11:54:23 AM GMT -05:00 US/Canada Eastern
>> Subject: Re: Using koan --replace-self problem w/ F10 and 11 - but not 9
>>
>>
>> Just a thought, check the distro entry cobbler import created. I had a
>> similar problem the other day and noticed that the import used
>> initrd-PAE.img vice initrd.img. Once I made the change normal PXE
>> installs started to work with F10.
>>
>> -Ron
>> Scot Floess wrote:
>>
>> All:
>>
>> I've tried to figure out this problem for the last 2 weeks...just
>> off and on when I had free time at home...
>>
>> I am running Cobbler 1.4.2 on CentOS 5.2...
>>
>> Basically this is the scenario: I downloaded the Fedora 10 DVD
>> ISO and imported it into Cobbler. I stepped onto 2 different
>> machines and tried to use koan --replace-self for the Fedora 10
>> distro I imported... Yes, I created a system, etc :)
>>
>> Anyway, I get a dialog box on both machines stating something
>> about trouble (sorry not sitting in front of the machines at the
>> moment) talking to the cobbler server - the URL displayed looks to
>> be correct. In looking at virtual console 2 or 3, I see the
>> following displayed:
>>
>> "no network devices in choose network device!"
>>
>> After some mucking around, I thought I'd try both Fedora 11 and 9
>> - basically doing the same process I did for 10 (download the ISO,
>> import, etc)...
>>
>> With Fedora 11 I got the same result as 10... For Fedora 9 - it
>> worked!
>>
>> I thought - OK, let me try using buildiso - see if, on the off
>> chance, it worked... Nope - same problem...
>>
>> Finally, I decided to grab the boot.iso from Fedora 10 - to see if
>> I could do an NFS install - that worked!
>>
>> At this point, I'm not sure what is causing this problem...or if
>> anyone has seen this issue? I had been using Cobbler 1.4.1 - but
>> jumped up to 1.4.2...
>>
>> Any and all help - is greatly appreciated!
>>
>> Flossy
>>
>>
>> --
>> Make It So Number One
>>
>> ------------------------------------------------------------------------
>>
>> _______________________________________________
>> cobbler mailing list
>> cobbler(a)lists.fedorahosted.org
>> https://fedorahosted.org/mailman/listinfo/cobbler
>>
>>
>>
>> _______________________________________________ cobbler mailing list
>> cobbler(a)lists.fedorahosted.org
>> https://fedorahosted.org/mailman/listinfo/cobbler
>>
>> --
>> Make It So Number One
>> ------------------------------------------------------------------------
>>
>> _______________________________________________
>> cobbler mailing list
>> cobbler(a)lists.fedorahosted.org
>> https://fedorahosted.org/mailman/listinfo/cobbler
>>
>>
>
> _______________________________________________
> cobbler mailing list
> cobbler(a)lists.fedorahosted.org
> https://fedorahosted.org/mailman/listinfo/cobbler
>
>
_______________________________________________
cobbler mailing list
cobbler(a)lists.fedorahosted.org
https://fedorahosted.org/mailman/listinfo/cobbler
--
Make It So Number One
_______________________________________________
cobbler mailing list
cobbler(a)lists.fedorahosted.org
https://fedorahosted.org/mailman/listinfo/cobbler
--
Make It So Number One
15 years, 2 months
Traffic ... and development list reconsidered
by Michael DeHaan
Seeing we've had like 50 mailing list messages today (and this is WAAY
too many for me to keep up with them), I'm going to be asking for a
cobbler-devel mailing list for development topics. I know some folks
don't want multiple lists, but it's time. I'll let you know when it is
available.
Patches will all go to the development list as will future release
planning. Also, the more people who can use github and such for
patches, the better. This makes merging things and keeping up with
patches easier.
I'll be looking for folks to help a bit more with the user Q&A on this
list and on IRC so we can keep the development topics moving forward.
Meanwhile, if you have an issue you definitely want my attention on,
please use the lists, rather than IRC... though IRC is a great place for
users to ask questions it's hard keeping up with all the scrollback.
Thanks!
--Michael
15 years, 2 months