SDB:KIWI-LTSP on SLED 10

Jump to: navigation, search
Icon-warning.png
Warning: KIWI-LTSP is developed on openSUSE 11.1, the prebuilt image used for SLED are actually openSUSE 11.1 images. Mixing SLES and SLED and these unofficial packages may void your support contract.

Installation

  • Install SLED 10 SP2
  • Add SLES 10 SP2 repository via YaST2 Installation sources module, this is where server packages will get installed.
  • Add one script (become root: su -)
 cd /usr/bin
 wget -c http://forgeftp.novell.com/kiwi-ltsp/ip-easy-ltsp-config.py
 chmod +x ip-easy-ltsp-config.py
  • Add openSUSE:Tools and server:ltsp repositories
 zypper sa http://download.opensuse.org/repositories/openSUSE:/Tools:/Devel/SLE_10/ openSUSE:Tools
 zypper sa http://download.opensuse.org/repositories/server:/ltsp/SLE_10/ server:ltsp
 zypper refresh
  • Install kiwi-ltsp-prebuilt
 zypper in kiwi kiwi-pxeboot kiwi-desc-ltsp kiwi-ltsp-prebuilt kiwi-ltsp-bootimages ltsp-server ltspfs

There may be complaints about missing squashfs and easy-ltsp, ignore and continue.

Configuration

  • Use "traditional method" for network configuration, give static IP 10.0.0.254 to the internal interface on the server.
  • Edit /etc/sysconfig/kiwi-ltsp "DHCP_Interface=ANY", change "ANY" to the interface your dumb clients are on.
  • Turn off firewall to the interface connected to internal network.
  • Run the following to configure all the services:
 kiwi-ltsp-setup -c

If you give any other IP address change /etc/sysconfig/kiwi-ltsp''' "SERVER_IP" variable accordingly.

Troubleshooting

  • If a client gives a "Could not find rootfs..." error during boot you need to remove the "COMBINED_IMAGE=" line from /srv/tftpboot/KIWI/config.default
  • If a clients screen starts blinking after most of the boot process is finished you need to change "CONFIGURE_X=False" to "CONFIGURE_X=True" in /srv/tftpboot/KIWI/lts.conf
  • If clients mysteriously can't find a DHCP server to start the TFTP boot process the dhcp daemon may have died. To check open a terminal and type "rcdhcpd status" if the output says "dead" issue an "rcdhcpd restart" I wrote a little bash script that issues "rcdhcpd restart" and put it in /etc/cron.hourly. I am not sure what is causing this issue but one hint might be in /var/log/warn where it reports "dhcpd: Can't create new lease file: Permission denied". Issue "less /var/log/warn | grep dhcpd" to check that.

See Also