written by Peter van der Walt for www.netcb.com

 

Introduction

Enabling the GWHA component on Linux enables your GroupWise system to automatically restart agents when they stop running for whatever reason. This greatly reduces manual administration and system downtime.

I have been configuring this on all my GroupWise installations on Linux and have only had great experiences with this.

I have found the Novell documentation to be a bit vague, so it took me a day or two to figure out how to actually deploy GWHA - I thus decided to create a article showing how easily it can be deployed, requiring only 5 simple steps!

Steps

1. Install GroupWise Monitor on one of the GroupWise servers
   
   
   
   1. Install and Configure GW Monitor on one server
   
   

• Edit the /etc/init.d/grpwise-ma script and uncomment the MA-OPTIONS line.In the screenshot below, you can see a Sample:
  
  
  
  Leave the Username as 'gwha', change the password to whatever you want to use, and change the Poll time to how many seconds you want to wait before checking the agents are still running.

• On all the GroupWise servers to be monitored:-- check /etc/opt/novell/groupwise/gwha.conf
  -- run 'useradd gwha' to create a local user for GWA to run as
  -- run 'passwd gwha' to set the user's password to the one used in step 2.
  -- Yast -> Xinetd -> gwha -> Toggle Status On - to ensure that the GWHA is enabled.

• Test that the GWHA daemon is listening (netstat -nl | grep 8400)

• Test by killing the agent's processes and confirm that GWHA restarts the agents
  
  Do a 'ps -eaf | grep mta' for example, and kill the MTA's process using a 'kill -9 <PID>'
  
  Do a 'rcgrpwise status' to see if the MTA's status goes "dead" followed by "running" after the GroupWise monitor's poll time passed, it detected the agent as dead and restarted it.

Conclusion

As simple as that! In 5 easy steps you have configured GroupWise High Availability!

If any of the gwpoa, gwmta, gwinter or gwia processes should stop running - the GroupWise monitor will detect the agent status as Dead and issue the start command through the GWHA daemon - in seconds without any operator intervention!