====== Server preparation - Server monitoring ======
Automatic monitoring of production system is crucial for bussiness continuity. Monitoring is recommended also for the testing environment, but it is not mandatory.
This page will show you how to set up basic monitoring of server with CzechIdM using Nagios NRPE. It is very useful to store monitored values for trend overview (e.g. with Munin). Some monitoring systems (like Zabbix) can store trends and monitor services at once. It is also practical to install ''iostat'', ''vmstat'' and ''sar'' utilities on the server.

<note>This article is about real-time monitoring of the server and its services. It does not deal with monitoring of "the insides" of CzechIdM.</note>

**Example server parameters for this guide**

Table with example monitoring parameters was created for these server resources.
  * RHEL7-flavoured system.
  * About 100GB HDD.
  * At least 8GB RAM.
  * At least 2x2GHz CPU.
When implementing server monitoring adjust monitored parameters for your particular deployment.

===== Monitored parameters =====
This is a list of monitored server's (and services') parameters. It should be treated as a bare minimum and, if needed, extended according to your company's policy.
Parameters and their thresholds mentioned below are based on our best practices for the monitoring of a deployment.

^Service/Parameter ^Probe binary ^Name in NRPE ^Warning threshold ^Critical threshold ^Check frequency ^Notification frequency ^
|HOST UP| N/A | this is not implemented on the target machine | N/A or ping RTT threshold | high ping RTT or host is not pingable at all | every 5 minutes | every 6 hours |
|swap used space | check\_swap | check\_swap | 50% swap free | 10% swap free | every 5 minutes | every 24 hours |
|disk free space | check\_disk | check\_disk | 90% used | 95% used | every 5 minutes | every 24 hours |
|system load | check\_load | check\_load | 4,3.5,3 | 6,5.5,5 | every 5 minutes | every 24 hours |
|used memory | check\_mem | check\_mem | 90% used | 95% used | every 5 minutes | every 24 hours |
|process count | check\_procs | check\_procs | 300+ | 500+ | every 5 minutes | every 24 hours |
|zombie process count | check\_procs | check\_zombies | 1+ | 5+ | every 5 minutes | every 24 hours |
|system time | check\_ntp\_time | check\_time | skew >1min | skew >5min | every hour | every 24 hours |
|CzechIdM is running | check\_http | check\_idm | N/A | CzechIdM not running | every 5 minutes | every 24 hours |
|HTTPD is running | check\_http | check\_httpd | response time >1s | HTTPD is not running | every 5 minutes | every 24 hours |
|HTTPS certificate expiration | check\_http | check\_httpd\_cert | less than 30 days | less than 7 days | once a day | every 24 hours |
|PostgresSQL is running | check\_pgsql | check\_postgres | response time >0.5s | response time >1s or not running at all | every 5 minutes | every 24 hours |

===== Implementation =====
We will use nrpe and probes from the standard system packages. We have epel repository enabled.
  * NRPE daemon will listen on 5666\tcp (its default port). Open the port in your iptables by adding the rule: ''-A INPUT -m state --state ESTABLISHED,RELATED -p tcp --dport 5666 -j ACCEPT''.
  * All probes are located in their default installation location ''/usr/lib64/nagios/plugins/''.
  * We use one external probe check\_mem which can be downloaded here: [[https://exchange.nagios.org/directory/Plugins/System-Metrics/Memory/check_mem-2Esh/details]]. This probe, however, returns bad results on RHEL7 because of the different meaning of the ''free'' command output. You can dowload the fixed version from [[https://github.com/bcvsolutions/czechidm-monitoring/blob/master/monitoring/nagios-plugins/check_mem/check_mem|here]].

**Deployment**

First, install necessary packages:
<code>
yum install nrpe nagios-plugins-nrpe nagios-plugins-swap nagios-plugins-disk nagios-plugins-load nagios-plugins-procs nagios-plugins-ntp nagios-plugins-http nagios-plugins-pgsql
</code>
If you use SELinux, we need to permit the check\_disk plugin access to the ''/sys/kernel/...''. Easiest way (but not necessarily the most correct) is to set permissive mode for some plugins:
<code>
yum install policycoreutils-python
semanage permissive -a nagios_checkdisk_plugin_t
</code>
Edit the ''/etc/nagios/nrpe.cfg'' file and add your monitoring server address to the allowed\_hosts directive:
<code>
allowed_hosts=127.0.0.1,IPofMonitoringServer
</code>
Create a configuration of system checks in the file ''/etc/nrpe.d/checks.cfg''. Fill in the ''YOUR\_NTP\_SERVER'' and ''IDM\_SERVICE\_DOMAIN\_NAME'' accordingly. The ''MONITORING\_USER'' and ''MONITORING\_USER\_PASSWORD'' are values filled with credentials of an user which is capable to log into the PostgreSQL database. **Create separate user just for this purpose**.
<file txt checks.cfg>
command[check_swap]=/usr/lib64/nagios/plugins/check_swap -w 50% -c 10%
command[check_disk]=/usr/lib64/nagios/plugins/check_disk -w 90 -c 95
command[check_load]=/usr/lib64/nagios/plugins/check_load -w 4,3.5,3 -c 6,5.5,5
command[check_mem]=/usr/lib64/nagios/plugins/check_mem -w 90 -c 95
command[check_procs]=/usr/lib64/nagios/plugins/check_procs -w 300 -c 500
command[check_zombies]=/usr/lib64/nagios/plugins/check_procs -w 1 -c 5 -s Z
command[check_time]=/usr/lib64/nagios/plugins/check_ntp_time -H YOUR_NTP_SERVER -w60 -c300
command[check_idm]=/usr/lib64/nagios/plugins/check_http -H 127.0.0.1 -p 8080 -u '/idm/api/v1/status'
command[check_httpd]=/usr/lib64/nagios/plugins/check_http -H IDM_SERVICE_DOMAIN_NAME -S -p443 -w1
command[check_httpd_cert]=/usr/lib64/nagios/plugins/check_http -H IDM_SERVICE_DOMAIN_NAME -S -p443 -C30,7
command[check_postgres]=/usr/lib64/nagios/plugins/check_pgsql -H 127.0.0.1 -P 5432 -d template1 -l MONITORING_USER -p MONITORING_USER_PASSWORD -w0.5 -c1
</file>
Add the ''check\_mem'' script to the ''/usr/lib64/nagios/plugins/'' directory, make it executable:
<code>
cp check_mem /usr/lib64/nagios/plugins/
chmod 755 /usr/lib64/nagios/plugins/check_mem
</code>
Create the ''MONITORING\_USER'' in the PostgreSQL. Generate some strong password - you can use ''pwgen'' for that.
<code>
create user monitoring password 'somepassword';
</code>
Start and enable the NRPE daemon:
<code>
systemctl start nrpe
systemctl enable nrpe
</code>
To test the probes, you can use ''check\_nrpe'' plugin:
<code>
/usr/lib64/nagios/plugins/check_nrpe -H 127.0.0.1 -b 127.0.0.1 -c check_swap
</code>

**Nagios server configuration**

This is a sample configuration for the Nagios server. It is meant more as an inspiration, feel free to adapt it to your Nagios deployment.

Configure the ''check\_nrpe'' command (you probably already have this in your Nagios configuration):
<code>
define command{
        command_name check_nrpe
        command_line /usr/lib64/nagios/plugins/check_nrpe -H $HOSTADDRESS$ -c $ARG1$
}
</code>

Define CzechIdM server host:
<code>
define host {
        use                             linux-server
        host_name                       czechidm_server
        alias                           idmserver.example.com - CzechIdM server
        address                         1.2.3.4
        check_period                    24x7
        # we expect interval_length=60 as is the default, so 1440*60s = 1 day
        notification_interval           1440
        notification_period             24x7
}
</code>

Define checks:
<code>
define service {
        use                             generic-service
        host_name                       czechidm_server
        service_description             SWAP
        check_command                   check_nrpe!check_swap
        # we expect interval_length=60 as is the default, so 5*60s = 5 minutes
        check_interval 			5
        # we expect interval_length=60 as is the default, so 1440*60s = 1 day
        notification_interval           1440
        contacts 			user1,user2
        contact_groups			admins1,admins2
}

... and similarly the other checks ...
</code>