StartCruiseControlServer

Kevin Brightwell edited this page Aug 26, 2015 · 1 revision
Clone this wiki locally

Introduction

This page shows you how to start / restart the CC server after a reboot (or if the service fails.

Run the following command on the build server (currently cruise.site.uottawa.ca) while logged in as the regular user of the server, not yourself:

~/bin/cc_start

If problems with cruise control persist, consider the following:

  • Look at the log output from cc_start to see if there are any clues as to the problem
  • Determine if any recently updated software (Java, etc.) might have resulted in the failure
  • Look at the cc/config.xml script and see if one of the 'on success' tasks is failing. These can be run manually to test that all is working OK. The tasks include gitrebase (to update the git repositories), promoteuol (to move UmpleOnline changes to the test mode), dumple (to update the diagram at http://metamodel.umple.org), bgmakejavadoc (to update the javadoc documentation for umple), and the ant target deployUmpleDocs to update the user manual.

Updating Java version

Download the Java rpm file from Oracle and put it in the tmp directory

As a superuser run sudo rpm -Uvh jdk-8

cd /h/ralph/umple_configs/src

If this is a major java release update /h/ralph/umple_configs/src to have jdk9 and for it to point to latest version in /usr/java/ Then Make sure /h/ralph/umple_configs/bin/cc_start is adjusted to accommodate

Run cc_start to ensure the latest is linked in

Future Work

The ~/bin/cc_start has been added to /etc/rc.d meaning that the cruise control server starts automatically. But we experienced an issue where the rake task was not accessible from this instance. This needs investigating. The path to the rake being used might not be correct when started by the init process, or there may be a permission problem.

We might want to set up an init.d daemon, which would be located at

/etc/init.d/cc

And, then registered to be auto started on reboot.

sudo "update-rc.d -f cc defaults 90 10"

We could also integrate a monitoring program like monit, which can also "restart" processes when they fail (i.e. apart from a reboot).