Spacewalk and RHEL6

Register RHEL6 client with Spacewalk

Reference:

• http://linuxinternetworks.com/register-clients-in-spacewalk/ 

Instructions

1. First set a couple of variables to help the following instructions work with more versions:
   

   BASE_ARCH=$(uname -i)
   BASE_REL=$(sed -e '/.*release /s///' -e 's/\..*//' /etc/redhat-release)

2. Install the Space walk client for your architecture:
   

   rpm -Uvh http://spacewalk.redhat.com/yum/1.7/RHEL/$BASE_REL/$BASE_ARCH/spacewalk-client-repo-1.7-5.el$BASE_REL.noarch.rpm
   wget --directory-prefix=/tmp ftp://mirror.optus.net/epel/$BASE_REL/$BASE_ARCH/epel-release*.noarch.rpm && rpm -Uvh /tmp/epel-release*.noarch.rpm

3. Install the client packages:
   

   yum install rhn-client-tools rhn-check rhn-setup rhnsd m2crypto yum-rhn-plugin

4. Register the client:
   

   rhnreg_ks --serverUrl=https://<fqn_spacewalk_server>/XMLRPC --activationkey=1-<activation_key_label>

Install Spacewalk on RHEL6

Links

• https://fedorahosted.org/spacewalk/wiki/HowToInstall

Instructions

wget rpm -Uvh http://spacewalk.redhat.com/yum/1.7/RHEL/6/x86_64/spacewalk-repo-1.7-5.el6.noarch.rpm

Fedora16 with horde

Amazing, I have got horde with a postgres backend to work on Fedora16.

Links:

• https://fedorahosted.org/spacewalk/wiki/PostgreSQLServerSetup  -- helped with installing postgres and setting the permissions for accessing the database.
• http://forum.parallels.com/showthread.php?t=74634 -- php.ini change to pickup modules.  I'm using x86_64 so I guess things went wrong with the 'lib64' path not being included?
• http://wiki.kolab.org/Configure_Horde_with_PostgreSQL -- example postgres configuration with horde.
• http://www.horde.org/apps/horde/docs/UPGRADING -- Horde Upgrade Instructions

Install Postgres:

yum install php-pgsql postgresql postgresql-server
chkconfig postgresql on
postgresql-setup initdb
service postgresql start
su - postgres
cp /usr/share/horde/scripts/sql/create.pgsql.sql .
vim create.pgsql.sql  -- (set a password for the horde user)
psql -d template1 -f create.pgsql.sql
(if all the tables and indexes don't fly up the screen as they are created then you have a problem.  For me it took awhile to get pg_hba.conf correct.  See supplementary points below.)
exit  -- (return to the root user)
psql -h 127.0.0.1 -p 5432 -U horde -d horde  -- (to test logging into the horde database as the horde user.)
rm /var/lib/pgsql/create.pgsql.sql

Supplementary Points:

• Horde is out of date in the Fedora16 repository!  I persisted thinking that things will be easier if I use the Fedora packages :-(
• http://your-server/horde/test.php -- is not working and the doco I have found so far is not relevant for activating the test feature.
• Turn off selinux as the provided policies are not good enough.  (Now that I've saved you some time maybe you can work on the selinux policies for me???)
• Add the following to /etc/php.ini and service httpd restart for fedora16_x86-64:
• extension_dir = "/usr/lib64/php/modules"
•  Add the following to /var/lib/pgsql/data/pg_hba.conf before the "local all all peer" line to allow the horde database user to login via a localhost TCP socket.  Don't forget to reload the database configuration service postgresql reload :

local horde horde md5
host  horde horde 127.0.0.1/8 md5
host  horde horde ::1/128 md5
local horde postgres  ident

• Clean out the postgres database to retry the horde initialisation:

su - postgres
psql
DROP DATABASE horde;
DROP USER horde;

rsync of block devices

How to clone virtual disks as logical volumes with resume?

Some clever folks are working on patches to "rsync" to allow the copying of block devices:  http://lists.samba.org/archive/rsync/2010-June/025164.html

Paul Whittaker fixed me up with just such a patched version of rsync so I can move my VM disks around my network more efficiently.  The only thing to note is /dev/mapper/<logical_volume> is a symbolic link and you must use /dev/<vg>/<lv> to access the real block device.  

The example below is me running an "rsync" check over a previously cloned volume that failed to cleanly complete when using my favourite "nc" (netcat) method.  A three minute verify is not bad for an 8GB virtual disk!

root@blackdouglas> rsync -vP --inplace --copy-devices jd:/dev/mapper/JD-vm_canobolissmith_ROOT /dev/mapper/BD1-vm_canobolissmith_ROOT
JD-vm_canobolissmith_ROOT
           0 100%    0.00kB/s    0:03:18 (xfer#1, to-check=0/1)

sent 741502 bytes  received 370826 bytes  3570.88 bytes/sec
total size is 0  speedup is 0.00
root@blackdouglas>