Monitor the status of Hudson slaves.
Visit a url like http://myslave:3141 to see whether a slave is running and how much memory it is using. Configure the port used by clicking Manage Hudson on the dashboard.
Monitoring with Nagios
You can monitor your slaves using Nagios and its remote monitoring tool NRPE. To do this on a Linux slave:
- Install Nagios on a monitoring server and NRPE on the slave. Instructions are on the Nagios site.
- Verify that you are monitoring basics like CPU usage and disk space by visiting the Nagios Service page.
- Edit your Nagios config file (typically /etc/nagios/nrpe.cfg). Add a line like the one below. If you have changed the default port used by the slave-status plugin, replace 3141 with your port number.
| command[check_slave]=/usr/lib/nagios/plugins/check_http -H localhost -p 3141 -r '<status>Running</status>'
- Check that running this command on the monitoring server gives output about HTTP:
| /usr/lib/nagios/plugins/check_nrpe -H [slave server name] -c check_slave
- Edit the per-server config file on the monitoring server (on Ubuntu this is in /etc/nagios3/conf.d). Add a section like the one below.
host_name [server name]
service_description Hudson Slave
- Restart Nagios (with a command like sudo /etc/init.d/nagios3 restart) and check you can see a Hudson Slave line for your server on the Service page.
Note this just monitors that your slave is up. You can use a more complex matching rule in the check_slave command, or write your own, if you also want to monitor memory usage.
- Use new extensibility tools to allow easy addition of more monitoring (threads used, garbage collection, and so on). See Defining a new extension point.
- Add functional tests. See Unit Test (misnamed).
Version 1.4 (10 February 2010)
- Update code for more recent Hudson.
Version 1.3 (15 May 2009)
- Fix bug: immediately after installation, port number was zero. Now defaulted to 3141.
- Fix bug: serialising Logger inadvertently.
Version 1.2 (23 April 2009)