NRPE analogue for NCPA
Posted: Mon Aug 02, 2021 10:00 am
Is there a direct analogue to NRPE's host-defined setup in NCPA?
I'm really digging NCPA so far, especially for managing my Windows systems, but something that was really handy in NRPE is defining a somewhat generic service then modifying what specific command that service definition used on the actual NRPE client. So I have NRPE configs based on host-OS that get deployed but never really have to touch my Nagios definitions.
NRPE Example:
Monitoring updates on hosts. Using the same "check_updates" command via NRPE executes a different script depending on RedHat-based vs Debian-based. This is despite the fact that Nagios thinks it's the exact same service/command being executed.
NCPA example:
I'm currently only using NCPA on my windows servers in my environment. For Windows below is an example of what specifically triggered this concern. Essentially NCPA registers the network metric as two different Ethernet devices (Ethernet vs Ethernet0) depending on virtualized Windows vs bare-metal Windows. This is one small example and I circumvented it by using the same service definition with two different versions (Ethernet and Ethernet0) but excluded from the machines that aren't relevant.
Code: Select all
command[COMMAND_NAME]=I'm really digging NCPA so far, especially for managing my Windows systems, but something that was really handy in NRPE is defining a somewhat generic service then modifying what specific command that service definition used on the actual NRPE client. So I have NRPE configs based on host-OS that get deployed but never really have to touch my Nagios definitions.
NRPE Example:
Monitoring updates on hosts. Using the same "check_updates" command via NRPE executes a different script depending on RedHat-based vs Debian-based. This is despite the fact that Nagios thinks it's the exact same service/command being executed.
Code: Select all
define service {
use delay-notification-service
hostgroup_name linux-servers
service_description Updates Available on Server
check_command check_nrpe_updates
servicegroups Updates
}
define command {
command_name check_nrpe_updates
command_line $USER1$/check_nrpe -H $HOSTADDRESS$ -t 30 -c "check_updates"
}
I'm currently only using NCPA on my windows servers in my environment. For Windows below is an example of what specifically triggered this concern. Essentially NCPA registers the network metric as two different Ethernet devices (Ethernet vs Ethernet0) depending on virtualized Windows vs bare-metal Windows. This is one small example and I circumvented it by using the same service definition with two different versions (Ethernet and Ethernet0) but excluded from the machines that aren't relevant.
Code: Select all
# TODO: How can we define what interface to check depending on the NCPA host checked?
# Win10 hosts may use Ethernet or Wireless, Windows Server uses Ethernet0
# Need something more like NRPE's command definitions file for NCPA
define service {
use ncpa-service
service_description Network Bandwidth Usage -- Ethernet
host_name !metecs-saveeam, !metecs-wkstn19
check_command check_ncpa!-t 'geBpBH2txclJKk8MFTer' -P 5693 --delta -M 'interface/Ethernet0'
}
define service {
use ncpa-service
service_description Network Bandwidth Usage -- Ethernet
host_name metecs-saveeam, metecs-wkstn19, !metecs-ad1, !metecs-ad2, !metecs-qb2, !metecs-veeam
check_command check_ncpa!-t 'geBpBH2txclJKk8MFTer' -P 5693 --delta -M 'interface/Ethernet'
}