Nagios

Introduction

Nagios is an open source monitoring tool. Its standard (Core) version is free for download and use with no real limitations, its premium (XI) version offers additional features, most notably a GUI interface with which to configure it. Configuring Nagios can be challenging at first, requiring edits to multiple config files to get new monitoring working, but once you're got the logic and the pattern understood it becomes quite flexible.

Nagios is centred around device polling (it can receive SNMP traps, but its a more advanced feature), and the presentation of state data. Though the first thing to appreciate is that Nagios doesn't actually do any monitoring, at its core it's a task scheduling and state management engine. It needs third party plugins, which do the actual monitoring a report back the state of the host you're monitoring to it. There are plugins provided out-of-the-box, which will probably achieve most (if not all) of what you want.

Terminology

• host - A host is any network device that you want to monitor, be it a server, router, switch, SAN; anything that has an IP address.
• hostgroup - This is a collection of similar devices that you want to apply similar monitoring too, a host can be in more than one hostgroup.
• plugin - A plugin is a monitoring module, built to monitor/interface with a specific device, application etc,. It hides Nagios from the specific's of whatever its interfacing with.
• command - A command is command line call of a plugin with one or more parameters, which defines how you might use a plugin to test a host.
• service - A service is something that you care about on a host, that you want to test (eg web server response, ping, disk space, CPU,

Useful Paths etc

Path	Description
/etc/nagios3/conf.d	Config files - anything in here is parsed as config, filenames are for your convenience and are irrelevant to Nagios
/etc/nagios-plugins/config	Plugin commands
/usr/lib/nagios/plugins	Plugin executables
nagios3 -v /etc/nagios3/nagios.cfg	Config check - do before a restart to check a new config makes sense
service nagios3 restart	Restart service (reloads config - will fail if config is invalid!)

Create SNMP Checks

Everything here creates various checks for my QNAP NAS, which I've used as an example.

Define OID's to Poll

Before you start you need to know what SNMP OID's you want to poll, and what their values should be. For common devices and metrics you can often get by with a Google search or two, but it doesn't take much for you to need to get a bit more involved.

When it comes to investigating what OID's you can poll for a specific device, your friend is GetIf.

Having downloaded the MIB and done some probing GetIf, I've decided I need to monitor the following OID's...

OID	Description	Example Return Data
.1.3.6.1.4.1.24681.1.2.6.0	System Temperature	41 C/105 F
.1.3.6.1.4.1.24681.1.2.17.1.6.1	System Volume 1 Status	Ready
.1.3.6.1.4.1.24681.1.2.17.1.5.1	System Volume 1 Space	1.74 TB
.1.3.6.1.4.1.24681.1.2.11.1.4.1	Physical Disk 1 Status	ready
.1.3.6.1.4.1.24681.1.2.11.1.7.1	Physical Disk 1 SMART Status	GOOD
.1.3.6.1.4.1.24681.1.2.11.1.3.1	Physical Disk 1 Temperature	35 C/95 F

Create Commands

Each type of check needs a command defined for it, which where the SNMP OID that will be checked is defined. Commands are are not specific to a particular host, so could be run against any system for which the check would be valid. There is some flexibility in that if you've certain checks that will be similar (eg checks for status of disk 1, disk 2 etc) then you can add arguments to the checks that can be defined later on.

I created a new file, called /etc/nagios3/conf.d/commands_qnap.cfg and added the following...

System Temperature

define command{
        command_name    check_qnap_sys_temp
        command_line    /usr/lib/nagios/plugins/check_snmp -H '$HOSTADDRESS$' -o .1.3.6.1.4.1.24681.1.2.6.0 -w 45 -c 55 -l Temp -u C
        }

• -H '$HOSTADDRESS$' - This is a standard wildcard for all check commands, Nagios substitutes the device's IP address
• -o .1.3.6.1.4.1.24681.1.2.6.0 - The SNMP OID being checked
  • .iso.org.dod.internet.private.enterprises.storage.storageSystem.SystemInfo.SystemTemperature.0
• -w 45 - The warning threshold
• -c 55 - The critical threshold
• -l Temp - A label for the check (appears in the check's Status Information column in Nagios display)
• -u C - The units of the metric being checked (appears in the check's Status Information column in Nagios display)

Volume Status

define command{
        command_name    check_qnap_sysvol_status
        command_line    /usr/lib/nagios/plugins/check_snmp -H '$HOSTADDRESS$' -o .1.3.6.1.4.1.24681.1.2.17.1.6.$ARG1$ -l "Volume Status" -r "Ready"
        }

• -o .1.3.6.1.4.1.24681.1.2.17.1.6.$ARG1$ - The SNMP OID being checked, $ARG1$ is used as a wildcard so that if I had more than one volume I could repeat the check for volume 1, 2 etc without creating a separate check command for each.
  • .iso.org.dod.internet.private.enterprises.storage.storageSystem.SystemInfo.SystemVolumeTable.SysVolumeEntry.SysVolumeStatus.$ARG1$
• -r "Ready" - The text expected back from the poll, anything else causes a critical error

Volume Space

define command{
        command_name    check_qnap_sysvol_space
        command_line    /usr/lib/nagios/plugins/check_snmp -H '$HOSTADDRESS$' -o .1.3.6.1.4.1.24681.1.2.17.1.5.$ARG1$ -w $ARG2$: -c $ARG3$: -l "Volume Space" -u GB
        }

• -o .1.3.6.1.4.1.24681.1.2.17.1.5.$ARG1$ - The SNMP OID being checked, as above $ARG1$ is used as a command parameter so that if I had more than one volume I could repeat the check for volume 1, 2 etc without creating a separate check command for each.
• -w $ARG2$: - The warning threshold, defining it as a command parameter allows me to alter the service threshold without altering the command definition. The trailing  : makes it a should be more than check rather than the normal should be less than check.
• -c $ARG2$: - The critical threshold, defining it as a command parameter allows me to alter the service threshold without altering the command definition. The trailing  : makes it a should be more than check rather than the normal should be less than check.
  • .iso.org.dod.internet.private.enterprises.storage.storageSystem.SystemInfo.SystemVolumeTable.SysVolumeEntry.SysVolumeFreeSize.$ARG1$

Disk Status

define command{
        command_name    check_qnap_disk_status
        command_line    /usr/lib/nagios/plugins/check_snmp -H '$HOSTADDRESS$' -o .1.3.6.1.4.1.24681.1.2.11.1.4.$ARG1$ -m /etc/nagios3/mibs/QNAP-NAS.mib -l "Disk Status" -r 0
        }

• -o .1.3.6.1.4.1.24681.1.2.11.1.7.$ARG1$ - The SNMP OID being checked, similar to above $ARG1$ is used as a command parameter so that I can create separate checks for the individual disks without creating a separate check command for each.
  • .iso.org.dod.internet.private.enterprises.storage.storageSystem.SystemInfo.SystemHdTable.HdEntry.HdStatus.$ARG1$
• -m /etc/nagios3/mibs/QNAP-NAS.mib - Path to the QNAP MIB file. The value returned is an integer, 0 for ready/good, a negative value for a fault. In order to translate the value (eg -9) to its actual meaning (eg rwError), Nagios needs access to the MIB file. You will need to download it from your NAS (from the Network Services | SNMP Settings page), and copy it to path indicated on your Nagios server.
• -r 0 - The data expected back from the poll, 0 maps to readyanything else causes a critical error

Disk SMART Status

define command{
        command_name    check_qnap_disk_smart_status
        command_line    /usr/lib/nagios/plugins/check_snmp -H '$HOSTADDRESS$' -o .1.3.6.1.4.1.24681.1.2.11.1.7.$ARG1$ -l "SMART Info State" -r "GOOD"
        }

• -o .1.3.6.1.4.1.24681.1.2.11.1.7.$ARG1$ - The SNMP OID being checked, similar to above $ARG1$ is used as a command parameter so that I can create separate checks for the individual disks without creating a separate check command for each.
  • .iso.org.dod.internet.private.enterprises.storage.storageSystem.SystemInfo.SystemHdTable.HdEntry.HdSmartInfo.$ARG1$
• -r "GOOD" - The text expected back from the poll, anything else causes a critical error

Disk Temperature

define command{
        command_name    check_qnap_disk_temp
        command_line    /usr/lib/nagios/plugins/check_snmp -H '$HOSTADDRESS$' -o .1.3.6.1.4.1.24681.1.2.11.1.3.$ARG1$ -w 45 -c 55 -l Temp -u C
        }

• -o .1.3.6.1.4.1.24681.1.2.11.1.3.$ARG1$ - The SNMP OID being checked, as above $ARG1$ is used as a command parameter so that I can create separate checks for the individual disks without creating a separate check command for each.
  • .iso.org.dod.internet.private.enterprises.storage.storageSystem.SystemInfo.SystemHdTable.HdEntry.HdTemperature.$ARG1$

Create Services

Services are used to define a generic check command within the bounds of a specific service. So for example, you could define two separate disk space checks, using the same command definition, but with different alerting thresholds depending on your requirements.

Services need to be defined with...

• hostgroup_name - The hostgroup defines which servers will have the service checks applied to it. For a host to be checked for the service it needs to be a member of the hostgroup, see Create Hostgroup for further info.
• service_description - A name for the service check, this is what is displayed in the Service field on the Nagios display
• check_command - The command (and its parameters, if any) to perform the check.

I created a new file, called /etc/nagios3/conf.d/services_qnap.cfg, in which to add service definitions, examples of which are below...

define service{
        use                     generic-service
        hostgroup_name          qnap-nas
        service_description     Temp Sys
        check_command           check_qnap_sys_temp
        }

define service{
        use                     generic-service
        hostgroup_name          qnap-nas
        service_description     Status SysVol 1
        check_command           check_qnap_sysvol_status!1
        }

• Note the !1 at the end of the command in order to pass a parameter of 1 (ie 1st volume) to the command

define service{
       use                     generic-service 
       hostgroup_name          qnap-nas
       service_description     Space SysVol 1
       check_command           check_qnap_sysvol_space!1!.5!.25
       }

• Note the !1!.5!.25 at the end of the command in order to pass parameters for volume 1, warning threshold of .5TB, and critical threshold of .25TB to the command

define service{
       use                     generic-service 
       hostgroup_name          qnap-nas
       service_description     Status Disk 1
       check_command           check_qnap_disk_status!1
       }

define service{
       use                     generic-service 
       hostgroup_name          qnap-nas
       service_description     SMART Disk 1
       check_command           check_qnap_disk_smart_status!1
       }

define service{
       use                     generic-service 
       hostgroup_name          qnap-nas
       service_description     Temp Disk 1
       check_command           check_qnap_disk_temp!1
       }

Create Hostgroup

The hostgroup definition allows you to group one or more hosts together, in order to have service checks run against them. So in the above I created services that would apply to hosts in the qnap-nas hostgroup. I can then add my NAS server to this hostgroup in order for it to be monitored (hostgroup definitions are normally found in /etc/nagios3/conf.d/hostgroups_nagios2.cfg

define hostgroup {
       hostgroup_name qnap-nas
               alias           QNAP NAS
               members         nas
       }

If I wanted to monitor more than one NAS I could just add further members (comma separated, no spaces). Note that any hosts specified in a hostgroup must themselves have a host definition (normally found in /etc/nagios3/conf.d/hosts.cfg, for example...

define host{
       use                     generic-host  
       host_name               nas
       alias                   NAS
       address                 192.168.1.200
       }

Check Tuning

It's unlikely that you really want everything checked every 5 mins, 24 hours a day. Some services might get a bit flaky in the middle of the night when there's maintenance tasks running, or don't warrant being checked so frequently.

In general its better to make such changes to generic templates, that can then be applied to one or more service checks. You can then edit changes centrally, rather than going round and updating services. Templates can be daisy chained so that subsequent templates override or add to config (see http://nagios.sourceforge.net/docs/3_0/objectinheritance.html for further info).

Check Frequency

For services that don't need to be checked as often as every 5 mins, create a new service template for the check internal, and apply to the appropriate services.

Create a new service template for the check interval in /etc/nagios3/conf.d/generic-service_nagios2.cfg, See example below, which changes the check interval to every 30 mins. Note that the template uses the normal generic-service, and then overrides the normal_check_interval.

# Service template for low frequency checks
define service {
       name                            low-freq-svc-template
       use                             generic-service
       normal_check_interval           30
       }

Update the service config for any services that you want to have the new check interval. Change the use config line to use the new template name, for example...

define service{
       use                     low-freq-svc-template
       host_name               wib1.domain.com
       service_description     Wibble Sys
       check_command           check_wib_svc
       }

Maintenance Windows / Hours of Service

For services that don't need to be checked 24x7, create a new time period, and a new service template for that time period, and apply to the appropriate services.

Define a new time period in /etc/nagios3/conf.d/timeperiods_nagios2.cfg, example excludes 02:00 - 03:00 hrs Sunday, and 02:30 - 03:00 hrs all other days (to not check/notify on a particular day, simply leave that day out of the config)...

define timeperiod{
        timeperiod_name wibblehours
        alias           Wibble Hours
        sunday          00:00-02:00,03:00-24:00
        monday          00:00-02:30,03:00-24:00
        tuesday         00:00-02:30,03:00-24:00
        wednesday       00:00-02:30,03:00-24:00
        thursday        00:00-02:30,03:00-24:00
        friday          00:00-02:30,03:00-24:00
        saturday        00:00-02:30,03:00-24:00
        }

Then create a new service template for the time-period in /etc/nagios3/conf.d/generic-service_nagios2.cfg, See example below. Note that the template uses the normal generic-service, and then overrides the check_period and notification_period settings.

# Service template for Wibble system services
define service {
        name                            wibble-svc-template
        use                             generic-service
        check_period                    wibblehours
        notification_period             wibblehours
        }

Finally, update the service config for any services that you want to have the new time period. Change the use config line to use the new template name, for example...

define service{
       use                     wibble-svc-template
       host_name               wib1.domain.com
       service_description     Wibble Sys
       check_command           check_wib_svc
       }

Ubuntu Software Updates Monitor

I've spend a fair amount of time faffing around to find a method of checking my Ubuntu servers for updates. The inbuilt check_apt doesn't give me the results as I see when logging in to a server, and other methods I've found don't seem to work at all.

Given that I only really care about my servers that run Ubuntu 10 LTS, I decided to knock up a quick script that makes use of the same mechanism that's used to generate the MotD that you see when you login to the console. As its a simple locally run script, you also need to have NRPE running as well.

1. Set-up the check script on all servers to be monitored
   1. In /usr/lib/nagios/plugins download the script
      • wget http://dl.sandfordit.com/scripts/check_apt_upgrade
   2. Make the file executable
      • chmod +x check_apt_upgrade
   3. Update the /etc/nagios/nrpe.cfg to include the check
      • Add command[check_apt_upgrade]=/usr/lib/nagios/plugins/check_apt_upgrade
   4. Restart the NRPE server service
      • service nagios-nrpe-server restart
2. Update the Nagios server to poll the check
   1. Add the section below to the appropriate service config file
   2. Check your Nagios config is valid
      • nagios3 -v /etc/nagios3/nagios.cfg
   3. Restart Nagios
      • service nagios3 restart

define service {
       hostgroup_name                  nrpe-std
       service_description             Updates
       check_command                   check_nrpe_port!check_apt_upgrade
       use                             generic-service
       notification_interval           0 ; set > 0 if you want to be renotified
}

SNMP Based (Michal Ludvig)

The check script that is called by SNMP doesn't work! I've left this here for the time being as the remote SNMP exec mechanism does work, and I expect to use it at some point. When I do, I'll remove this, and document that instead.

This check uses some scripts developed by Michal Ludvig, I've downloaded the scripts to my site, but the originals, complete with his supporting notes can be found here - http://www.logix.cz/michal/devel/nagios. Though note that I've updated the check_snmp_extend.sh script (didn't work for me, suspect Nagios file locations have changed since script was originally written), all kudos should still go to Michal.

To summarise how it works...

1. Nagios uses a local script to SNMP query a remote server you want to check
2. The SNMP query triggers another script to be run on the remote server which queries whether there are any updates to install
3. The result is returned via SNMP to the calling script on the Nagios server, which in turn passes the data to Nagios

To set it up...

1. On your Nagios server...
   1. Download check_snmp_exec.sh to /usr/lib/nagios/plugins
      • EG when in folder /usr/lib/nagios/plugins do wget http://dl.sandfordit.com/scripts/check_snmp_extend.sh
   2. Make the the script executable
      • EG chmod +x check_snmp_extend.sh
   3. Define a command for the check in /etc/nagios3/conf.d/commands.cfg (see below - Nagios command)
2. On your monitored servers (do one 1st to test)...
   1. Download check-apt-upgrade.pl to /usr/local/bin/
      • EG when in folder /usr/local/bin/ do wget http://dl.sandfordit.com/scripts/check-apt-upgrade.pl
   2. Make the the script executable
      • EG chmod +x check-apt-upgrade.pl
   3. Make the server's SNMP daemon aware of it, edit /etc/snmp/snmpd.conf , add the following
      • extend sw-updates /usr/local/bin/check-apt-upgrade.pl --run
   4. Restart the SNMP daemon
      • service snmpd restart
3. Back on the Nagios server...
   1. Define a service for the check in /etc/nagios3/conf.d/services_nagios2.cfg (see below - Nagios service)
   2. Check your Nagios config is valid
      • nagios3 -v /etc/nagios3/nagios.cfg
   3. Restart Nagios
      • service nagios3 restart

• nagios command...

define command{
       command_name    check_snmp_extend
       command_line    /usr/lib/nagios/plugins/check_snmp_extend.sh $HOSTADDRESS$ $ARG1$
}

• nagios service...

1. SNMP check for Ubuntu server package updates

define service {
       hostgroup_name                  ubuntu-servers
       service_description             Updates SNMP
       check_command                   check_snmp_extend!sw-updates
       use                             generic-service
       notification_interval           0 ; set > 0 if you want to be renotified
}

NRPE

The Nagios Remote Plugin Executor allows Nagios checks to completed on remote servers in a similar fashion to performing checks on the Nagios server. Whilst its not always necessary, as many remote checks can be performed by probing remotely accessible services (such as SNMP or HTTP), there are times when such checks are not suitable, for example...

• Running checks that aren't easily achievable via SNMP
• Checking local services such as MySQL that aren't accessible remotely from the server
• Running HTTP checks to test your web servers from more than one location
  • EG local to server to ensure the web-server itself is OK, and remotely to check that access is likely to OK for global users

The NRPE server that runs on remote monitored machines does require quite a few additional packages to be installed (see below for in-exhaustive list), and if you are concerned you can try the alternative approach of getting data back from your remote server via SNMP as described in this example Ubuntu Software Updates Monitor. This can make for a more lightweight solution, but will require you to write your own monitoring scripts to be called by the SNMP daemon.

Additional packages required by NRPE...

• mysql-common
• radiusclient1
• samba-common
• smbclient
• snmp

Setup

The procedures below will get NRPE running to monitor disk space, load and MySQL service availability on a remote server.

1. Install the NRPE Plugin on the main Nagios server
   • apt-get install nagios-nrpe-plugin
2. Install the NRPE Server on the remote/monitored server
   • apt-get install nagios-nrpe-server
3. On the remote/monitored server update the config /etc/nagios/nrpe.cfg for
   • Nagios communications...
     • EG server_port=5700 change the port if your monitored server is on the internet
     • EG allowed_hosts=192.168.1.25 change to the address of your Nagios server
   • Checks (some may already exist in config)...
     • Load: command[check_load]=/usr/lib/nagios/plugins/check_load -w 15,10,5 -c 30,25,20
     • Disk space: command[check_disks]=/usr/lib/nagios/plugins/check_disk -w 20% -c 10% -p /dev/mapper/svr-root
       • The disk path must be valid, do a df -h and update /dev/mapper/svr-root as required
     • MySQL: command[check_mysql]=/usr/lib/nagios/plugins/check_mysql -H 127.0.0.1 -u nagios -p poller
       • Assumes you have added the nagios user to MySQL, EG mysql -u root -p -e "create user nagios identified by 'poller';"
4. Restart service on remote/monitored server to apply config
   • service nagios-nrpe-server restart
5. If necessary (if you're using the server's firewall), open the NRPE port on the remote/monitored server
   • ufw allow proto tcp from 192.168.1.25 to any port 5700
6. On the Nagios server create a hostgoup for the checks (see Nagios Hostgroup below)
   • Edit hostgroups_nagios2.cfg file
7. On the Nagios server create a custom NRPE command (see Nagios Command below)
   • Edit commands.cfg file
8. On the Nagios server create a NRPE service file (see NRPE Services below)
   • Edit services_nrpe.cfg file
9. On the Nagios server, validate the config, and assuming all OK, restart service to apply
   • nagios3 -v /etc/nagios3/nagios.cfg
   • service nagios3 restart

• Nagios Hostgroup ( hostgroups_nagios2.cfg )

define hostgroup {
       hostgroup_name          nrpe-std
               alias           NRPE Standard servers
               members         wiki
       }

• Nagios Command ( commands.cfg )

define command {
       command_name    check_nrpe_port
       command_line    /usr/lib/nagios/plugins/check_nrpe -H $HOSTADDRESS$ -p 5700 -c $ARG1$
}

• NRPE Services ( services_nrpe.cfg )

# NRPE Standard checks

define service {
       hostgroup_name                  nrpe-std
       service_description             Load
       check_command                   check_nrpe_port!check_load
       use                             generic-service
       notification_interval           0 ; set > 0 if you want to be renotified
}

define service {
       hostgroup_name                  nrpe-std
       service_description             Disk Space
       check_command                   check_nrpe_port!check_disks
       use                             generic-service
       notification_interval           0 ; set > 0 if you want to be renotified
}

define service {
       hostgroup_name                  nrpe-std
       service_description             MySQL
       check_command                   check_nrpe_port!check_mysql
       use                             generic-service
       notification_interval           0 ; set > 0 if you want to be renotified
}

Web Site Content and Response Time Monitoring

The stock check_http is very good at basic web server checks, but once you host multiple sites, or want to monitor that your site is actually returning good pages it starts to lack. There are also plenty of plugins available for monitoring websites from the Nagios Plugin exchange. However none seemed to match the following requirements (I may have missed one that did)...

• Page content checking
  • This makes it possible to verify that the whole LAMP stack is working as expected.
• Web page response time
  • Important as just because your site(s) are delivering good content, if it takes over 3 secs, nobody is going to hang around to look at it
• Ability to monitor user/password protected pages

Therefore I took one that almost did, check_http_content, and modified it to match my requirements (which I'll upload to the exchange once I've got it working with the Nagios::Plugin Perl module), and called it check_url_content (for the time being its available via the previous link).

Script Options

Option	Purpose	Default
-U <url>	URL to retrieve (http or https)
-m <text>	Text to match in the output of the URL
-w <secs>	Warning time threshold	3 secs
-c <secs>	Critical time threshold	10 secs
-t <secs>	Timeout in seconds to wait for the URL to load	30 secs
-u <user>	Username (only required if server requires authentication)
-p <pass>	Password (required if Username specified)
-r <realm>	Realm (required if Username specified), when accessing a protected site an Authentication pop-up will display 'The site says "realm"'.
-h <host>	Host (optional when Username specified), should be in the following format 'www.domain.com:443'

Examples

• Basic check example

define command{
       command_name    check_http_content
       command_line    /usr/lib/nagios/plugins/check_http_content -U $ARG1$ -m $ARG2$
       }

define service{
       use                     generic-service
       host_name          	www.sandfordit.com
       service_description     www - vWiki
       check_command           check_http_content!http://www.sandfordit.com/vwiki!'Monitoring system'
       }

• All options example

define command{
       command_name    check_url_content_opt
       command_line    /usr/lib/nagios/plugins/check_url_content -U $ARG1$ -m $ARG2$ -r $ARG3$ -u $ARG4$ -p $ARG5$ -w $ARG6$ -c $ARG7$ -t $ARG8$
       }

define service{
       use                     generic-service
       host_name               www.sandfordit.com
       service_description     www - Secure
       check_command           check_url_content!'http://www.sandfordit.com/secure'!'This page works'!Realm!user!password!5!10!60
       }