Building Linux Terminal Service Cluster using LVS

Configuration Example

The basic issue faced by me during LVS configuration using piranha was that Nanny could not monitor the service and would frequently give a READ TIME OUT.

The configuration otherwise was a standard LVS-NAT configuration in rr mode. Once the connection with a Terminal Server was established, there would be direct communication with the selected Terminal Server.

Environment ->

LVS-Piranha Server(s) -> RHEL 5 (2.6.18-8.el5) running piranha (2 in number) Real Servers -> RHEL 4 (2.6.9-42.ELSmp) acting as Terminal Servers (xdm->gdm)

To get around the Nanny read time out, issue ->

On the Real servers ->

a) Enable remote shell access by the LVS server(s)

                  For RHEL 4 -> install rsh-server rpm
                  edit /etc/securetty
                  enable rsh and rlogin
                  restart xinetd
                  create .rhosts file in the home folder of the user (in my case root) to allow remote access from                   LVS Server(s)

b) Create a small script in /usr/local/bin (let's say xdmstatus) ->

                 #!/bin/bash

                  RESULT=`netstat -an | grep 177 | grep udp | awk '{ print $1 }'`
                  echo $RESULT

c) make this script executable (chmod a+x)

On the LVS Server(s) ->

modify /etc/sysconfig/ha/lvs.cf to contain (please see snippet) ->

             send_program = "/usr/bin/rsh %h /usr/local/bin/xdmstatus"
             expect = "udp"

External Links

• http://www.ltsp.org/

"Building Linux Terminal Service Cluster using LVS" is an LVS Example related stub. You can help LVSKB by expanding it