[cobbler] A question about pxe_just_once logic
Hi all,
I was building a CentOS 6.2 host that supports 9 CentOS 6.2 KVM guests. One of the guest
is the management node that runs cobbler 2.2.2.
I have always been using the pxe_just_once in /etc/cobbler/settings, but it's been a
while since I have touched cobbler, and I got into troubles initially, described below.
Given the way we do, I have never used koan to setup virtual machines. I just use a
simple Python script to parse a guest XML template file, and creates all necessary guest
XML files and a shell script to run the corresponding the 'cobbler system add
--name=...' commands. We do this since we want to use the same way as for physical
nodes.
But this time, I got in troubles PXE booting the 1st guest immediately.
Checking it further, I found in the management node's /var/lib/tftpboot/pxeboot.cfg,
the PXE boot file shows "dynamic" content, like this:
[...]
ks=http://192.168.123.3/cblr/svc/op/ks/system/vc1
[...]
AFAIR, it should be like this:
DEFAULT local
PROMPT 0
TIMEOUT 0
TOTALTIMEOUT 0
ONTIMEOUT local
LABEL local
LOCALBOOT -1
Puzzled, I thought it for a while, and decided to do the following:
cobbler system edit --name=vc1 --netboot-enabled=0
and then cobbler sync
Lo and behold, the corresponding PXE conf file now became correct.
I then did a
cobbler system edit --name=vc1 --netboot-enabled=1
and on the host, issued a
sudo virsh start vc1
followed by a
sudo virt-viewer vc1
Everything went as anticipated afterwards.
Sure, I got around the issue, but IMHO this part is confusing.
I think that it's just unreasonable to use the sequence (netboot-enabled=0; cobbler
sync; netboot-enabled=1) that I did above to enable a freshly defined node to PXE boot.
There should be a third state just for a fresh install.
Or, did I misunderstand anything?
Regards,
Zack