Starting in version 2.3.0, you can restart Puppet Server by sending a hangup signal, also known as a HUP signal or SIGHUP, to the running Puppet Server process. The HUP signal stops Puppet Server and reloads it gracefully, without terminating the JVM process. This is generally much faster than completely stopping and restarting the process. This allows you to quickly load changes to your Puppet Server master, including configuration changes.
There are several ways to send a HUP signal to the Puppet Server process, but the most straightforward is to run the following
kill -HUP `pgrep -f puppet-server`
Starting in version 2.7.0, you can also reload Puppet Server by running the “reload” action via the operating system’s service framework. This is analogous to sending a hangup signal but with the benefit of having the “reload” command pause until the server has been completely reloaded, similar to how the “restart” command pauses until the service process has been fully restarted. Advantages to using the “reload” action as opposed to just sending a HUP signal include:
Unlike with the HUP signal approach, you do not have to determine the process ID of the puppetserver process to be reloaded.
When using the HUP signal with an automated script (or Puppet code), it is possible that any additional commands in the script might behave improperly if performed while the server is still reloading. With the “reload” command, though, the server should be up and using its latest configuration before any subsequent script commands are performed.
Even if the server fails to reload and shuts down — for example, due to a configuration error — the
kill -HUP command might still return a 0 (success) exit code. With the “reload” command, however, any configuration change which causes the server to shut down will produce a non-0 (failure) exit code. The “reload” command, therefore, would allow you to more reliably determine if the server failed to reload properly.
Use the following commands to perform the “reload” action for Puppet Server.
All current OS distributions:
service puppetserver reload
OS distributions which use sysvinit-style scripts:
OS distributions which use systemd service configurations:
systemctl reload puppetserver
Note: If you’re using Puppet Enterprise (PE), you can reload the server from the command line by running
service pe-puppetserver reload. However if you need to change a setting, do so in console or with Heira, and then the agent will reload the server when it applies the change. For more information, see Configuring and tuning Puppet Server.
There are three ways to trigger your Puppet Server environment to refresh and pick up changes you’ve made. A request to the HTTP Admin API to flush the JRuby pool is the quickest, but picks up only certain types of changes. A HUP signal or service reload is also quick, and applies additional changes. Other changes require a full Puppet Server restart.
Note: Changes to Puppet Server’s logging configuration in
logback.xmldon’t require a server restart. Puppet Server recognizes and applies them automatically, though it can take a minute or so for this to happen. However, you can restart the service to force it to recognize those changes.
hiera.yamlfile to change your Hiera configuration.
puppet cert clean
puppet cert clean
ca.cfgto enable or disable Puppet Server’s certificate authority (CA) service.
For these types of changes, you must restart the process by using the operating system’s service framework, for example, by using the
Note: To ensure that the Puppet master and CA service is running in a platform agnostic way, use the
puppet resource service puppetserver ensure=runningcommand. This command is equivalent to
systemctl start puppetserveron systems that support it. For more information on the resource command and managing a server’s desired state, see Man Page: puppet resource and Resource Type: service.