cobbler suddenly started putting IP addresses into the pxelinux config
by Matthew Miller
So, our cobbler server was going along fine, doing the right thing, but
suddenly it has started putting the IP address of the cobbler web server
from which to get the kickstart file into the per-system pxelinux config
files instead of using the name (as given in the system config file).
This seems to coincide with the update to the new 2.0.9 version in EPEL.
This breaks our setup, because we're actually behind a proxy that uses
named-based virtual hosts.
If one wanted to use IP addresses rather than names, we could put *those* in
the config file. Why the change? Is there an option somewhere to force the
old behavior?
Thanks!
--
Matthew Miller mattdm(a)mattdm.org <http://mattdm.org/>
13 years, 2 months
Apache httpd warning about FQDN
by petermity
After setting up Cobbler 2.0.10, during system startup, or if I
restart httpd manually, I get:
Starting httpd: httpd: Could not reliably determine the server's
fully qualified domain name, using 127.0.0.1 for ServerName
I fixed it manually by setting /etc/httpd/conf/httpd.conf
ServerName to the IP of eth0, but is this something that
Cobbler's system setup could take care of, or that "cobbler
check" should at least report? (I didn't notice it at first).
This PC is dedicated to Cobbler use and is on no network (just
connected to a switch to build targets).
13 years, 2 months
need to re-run cobbler aclsetup after all cobblerd restarts
by Tom Georgoulias
I built a new cobbler server using cobbler-2.0.10-1.el5.noarch.rpm from
EPEL-Testing and noticed that the ACLs have to be redone every time that
cobblerd is restarted. Can anyone else confirm this? I observed this
behavior on 2.0.9 as well, but not on 1.6.6 (which was the release I was
using previously).
Here's the process I'm using:
As root, grant permissions to my dept group name:
# cobbler aclsetup --addgroup=GROUPNAME
As a user that group:
$ cobbler
usage
=====
cobbler <distro|profile|system|repo|image> ...
[add|edit|copy|getks*|list|remove|rename|report] [options|--help]
cobbler
<aclsetup|buildiso|deploy|import|list|report|reposync|sync|validateks|version>
[options|--help]
As root, restart cobblerd:
# service cobberd restart
After the restart,
$ cobbler
User cannot run command line, need read access to /var/lib/cobbler/web.ss
As root, regrant the permissions:
# cobbler aclsetup --addgroup=GROUPNAME
After this, everything works.
Tom
13 years, 2 months
.link_cache warning on cobbler sync
by petermity
Every time I do "cobbler sync", as part of the output I get
this:
cleaning link caches
running: find /var/lib/tftpboot/.link_cache -maxdepth 1 -type f -links 1
-exec rm -f '{}' ';'
received on stdout:
recieved on stderr: find: `/var/lib/tftpboot/.link_cache': No such file or
directory
running: find /var/www/cobbler/.link_cache -maxdepth 1 -type f -links 1
-exec rm -f '{}' ';'
received on stdout:
recieved on stderr: find: `/var/www/cobbler/.link_cache': No such file or
directory
The sync seems to work and reports no other problems.
What do I need to do to fix this so that there are no warnings or
errors when a sync is done?
13 years, 2 months
MemTest bug still exists - why?
by petermity
When I first tried Cobbler, choosing MemTest in the PXE menu just
flickered and did nothing. I quickly realized it's because its
image is in the wrong place on the server.
Searching the mailing list, this was discussed, and a patch
suggested, over 2 years ago. Over 1 year ago, Trac issue #574
was entered for it.
Is there a reason it hasn't been fixed yet?
For now, I fixed it manually. Actually, fixing it manually, it
gets broken again at the next "cobbler sync". To work around it,
I created file /var/lib/cobbler/triggers/sync/post/fix-memtest
with content:
#!/bin/bash
cp /var/lib/tftpboot/memtest* /var/lib/tftpboot/images
and did chmod +x on it. Now MemTest works after "cobbler sync".
13 years, 2 months