Monitoring cluster nodes

Manually checking cluster status

GitHub Enterprise Server has a built-in command line utility for monitoring the health of the cluster. From the administrative shell, running the ghe-cluster-status command executes a series of health checks on each node including verification of connectivity and service status. The output shows all test results including the text ok or error. For example, to only display failing tests, run:

admin@ghe-data-node-0:~$ ghe-cluster-status | grep error
> mysql-replication ghe-data-node-0: error Stopped
> mysql cluster: error

Monitoring cluster status with Nagios

You can configure Nagios to monitor GitHub Enterprise Server. In addition to monitoring basic connectivity to each of the cluster nodes, you can check the cluster status by configuring Nagios to use the ghe-cluster-status -n command. This returns output in a format that Nagios understands.

Prerequisites

• Linux host running Nagios.
• Network access to the GitHub Enterprise Server cluster.

Configuring the Nagios host

1. Generate an SSH key with a blank passphrase. Nagios uses this to authenticate to the GitHub Enterprise Server cluster.

   nagiosuser@nagios:~$ ssh-keygen -t rsa -b 4096
   > Generating public/private rsa key pair.
   > Enter file in which to save the key (/home/nagiosuser/.ssh/id_rsa):
   > Enter passphrase (empty for no passphrase): leave blank by pressing enter
   > Enter same passphrase again: press enter again
   > Your identification has been saved in /home/nagiosuser/.ssh/id_rsa.
   > Your public key has been saved in /home/nagiosuser/.ssh/id_rsa.pub.

   Security Warning: An SSH key without a passphrase can pose a security risk if authorized for full access to a host. Limit this key's authorization to a single read-only command.

2. Copy the private key (id_rsa) to the nagios home folder and set the appropriate ownership.

   nagiosuser@nagios:~$ sudo cp .ssh/id_rsa /var/lib/nagios/.ssh/
   nagiosuser@nagios:~$ sudo chown nagios:nagios /var/lib/nagios/.ssh/id_rsa

3. To authorize the public key to run only the ghe-cluster-status -n command, use a command= prefix in the /data/user/common/authorized_keys file. From the administrative shell on any node, modify this file to add the public key generated in step 1. For example: command="/usr/local/bin/ghe-cluster-status -n" ssh-rsa AAAA....

4. Validate and copy the configuration to each node in the cluster by running ghe-cluster-config-apply on the node where you modified the /data/user/common/authorized_keys file.

   admin@ghe-data-node-0:~$ ghe-cluster-config-apply
   > Validating configuration
   > ...
   > Finished cluster configuration

5. To test that the Nagios plugin can successfully execute the command, run it interactively from Nagios host.

   nagiosuser@nagios:~$ /usr/lib/nagios/plugins/check_by_ssh -l admin -p 122 -H hostname -C "ghe-cluster-status -n" -t 30
   > OK - No errors detected

6. Create a command definition in your Nagios configuration.

   Example definition

   define command {
         command_name    check_ssh_ghe_cluster
         command_line    $USER1$/check_by_ssh -H $HOSTADDRESS$ -C "ghe-cluster-status -n" -l admin -p 122 -t 30
   }
   

7. Add this command to a service definition for a node in the GitHub Enterprise Server cluster.

   Example definition

   define host{
         use                     generic-host
         host_name               ghe-data-node-0
         alias                   ghe-data-node-0
         address                 10.11.17.180
         }
   
   define service{
           use                             generic-service
           host_name                       ghe-data-node-0
           service_description             GitHub Cluster Status
           check_command                   check_ssh_ghe_cluster
           }
   

Once you add the definition to Nagios, the service check executes according to your configuration. You should be able to see the newly configured service in the Nagios web interface.