Compare commits

..

No commits in common. "ad2f04e6a053335aed31fcdb8eb48ab1490c51b5" and "bdb0583395a2873a5571bf83207ae5c127362ef3" have entirely different histories.

9 changed files with 158 additions and 152 deletions

View File

@ -12,7 +12,7 @@ services:
build: build:
dockerfile: Dockerfile dockerfile: Dockerfile
container_name: nagios container_name: nagios
image: docker-registry1.in.thelinuxpro.net:5000/tlp/nagios:241230.0.1 image: docker-registry1.in.thelinuxpro.net:5000/tlp/nagios:241228.0.3
networks: networks:
infra_dev_net: infra_dev_net:
ipv4_address: 10.99.23.36 ipv4_address: 10.99.23.36

View File

@ -1,28 +0,0 @@
### Hosts
define host{
use docker-host
host_name pi501.in.thelinuxpro.net
alias pi501
address 10.99.22.51
}
define host{
use docker-host
host_name pi502.in.thelinuxpro.net
alias pi502
address 10.99.22.52
}
define host{
use docker-host
host_name pi503.in.thelinuxpro.net
alias pi503
address 10.99.22.53
}
define host{
use docker-host
host_name pi504.in.thelinuxpro.net
alias pi504
address 10.99.22.54
}

View File

@ -1,14 +0,0 @@
### Hosts
define host{
use web-host
host_name kameronkenny.com
alias kameronkenny.com
address 204.8.15.81
}
define host{
use web-host
host_name thelinux.pro
alias thelinux.pro
address 204.8.15.81
}

View File

@ -1,43 +0,0 @@
### Service definitions for all docker hosts
define service{
use generic-service,graphed-service
hostgroup_name docker-hosts
service_description PING
check_command check_ping!100.0,20%!500.0,60%
}
define service{
use generic-service,graphed-service
hostgroup_name docker-hosts
service_description Docker Daemon
check_command check_nrpe!check_proc_docker
}
define service{
use generic-service,graphed-service
hostgroup_name docker-hosts
service_description Disk: /
check_command check_nrpe!check_disk
}
define service{
use generic-service,graphed-service
hostgroup_name docker-hosts
service_description Disk: /var/lib/docker/volumes
check_command check_nrpe!check_disk_docker
}
define service{
use generic-service,graphed-service
hostgroup_name docker-hosts
service_description System: Load
check_command check_nrpe!check_load
}
define service{
use generic-service,graphed-service
hostgroup_name docker-hosts
service_description System: Total Processes
check_command check_nrpe!check_procs
}

View File

@ -1,14 +0,0 @@
### Service definitions for all web hosts
define service{
use generic-service,graphed-service
hostgroup_name web-hosts
service_description PING
check_command check_ping!100.0,20%!500.0,60%
}
define service{
use generic-service,graphed-service
hostgroup_name web-hosts
service_description HTTPS
check_command check_https
}

View File

@ -4,7 +4,7 @@
# #
# NOTES: This config file provides you with some example command definitions # NOTES: This config file provides you with some example command definitions
# that you can reference in host, service, and contact definitions. # that you can reference in host, service, and contact definitions.
# #
# You don't need to keep commands in a separate file from your other # You don't need to keep commands in a separate file from your other
# object definitions. This has been done just to make things easier to # object definitions. This has been done just to make things easier to
# understand. # understand.
@ -17,7 +17,7 @@
# SAMPLE NOTIFICATION COMMANDS # SAMPLE NOTIFICATION COMMANDS
# #
# These are some example notification commands. They may or may not work on # These are some example notification commands. They may or may not work on
# your system without modification. As an example, some systems will require # your system without modification. As an example, some systems will require
# you to use "/usr/bin/mailx" instead of "/usr/bin/mail" in the commands below. # you to use "/usr/bin/mailx" instead of "/usr/bin/mail" in the commands below.
# #
################################################################################ ################################################################################
@ -47,7 +47,7 @@ define command{
# This command checks to see if a host is "alive" by pinging it # This command checks to see if a host is "alive" by pinging it
# The check must result in a 100% packet loss or 5 second (5000ms) round trip # The check must result in a 100% packet loss or 5 second (5000ms) round trip
# average time to produce a critical error. # average time to produce a critical error.
# Note: Five ICMP echo packets are sent (determined by the '-p 5' argument) # Note: Five ICMP echo packets are sent (determined by the '-p 5' argument)
@ -65,7 +65,7 @@ define command{
# SAMPLE SERVICE CHECK COMMANDS # SAMPLE SERVICE CHECK COMMANDS
# #
# These are some example service check commands. They may or may not work on # These are some example service check commands. They may or may not work on
# your system, as they must be modified for your plugins. See the HTML # your system, as they must be modified for your plugins. See the HTML
# documentation on the plugins for examples of how to configure command definitions. # documentation on the plugins for examples of how to configure command definitions.
# #
# NOTE: The following 'check_local_...' functions are designed to monitor # NOTE: The following 'check_local_...' functions are designed to monitor
@ -143,14 +143,9 @@ define command{
# 'check_http' command definition # 'check_http' command definition
define command{ define command{
command_name check_http command_name check_http
command_line $USER1$/check_http -I $HOSTADDRESS$ -H $HOSTNAME$ command_line $USER1$/check_http -I $HOSTADDRESS$ $ARG1$
} }
# 'check_https' command definition
define command{
command_name check_https
command_line $USER1$/check_http -S -I $HOSTADDRESS$ -H $HOSTNAME$
}
# 'check_ssh' command definition # 'check_ssh' command definition
define command{ define command{
@ -236,7 +231,7 @@ define command {
# #
# These are sample performance data commands that can be used to send performance # These are sample performance data commands that can be used to send performance
# data output to two text files (one for hosts, another for services). If you # data output to two text files (one for hosts, another for services). If you
# plan on simply writing performance data out to a file, consider using the # plan on simply writing performance data out to a file, consider using the
# host_perfdata_file and service_perfdata_file options in the main config file. # host_perfdata_file and service_perfdata_file options in the main config file.
# #
################################################################################ ################################################################################

View File

@ -1,3 +1,114 @@
### Host Template
define host{
name docker-host
use generic-host
hostgroups docker-hosts
check_period 24x7
check_interval 5
retry_interval 1
max_check_attempts 3
check_command check-host-alive
notification_period 24x7
notification_interval 120
notification_options d,u,r
contact_groups admins
register 0
}
### Hosts
define host{
use docker-host
host_name pi501.in.thelinuxpro.net
alias pi501
address 10.99.22.51
}
define host{
use docker-host
host_name pi502.in.thelinuxpro.net
alias pi502
address 10.99.22.52
}
define host{
use docker-host
host_name pi503.in.thelinuxpro.net
alias pi503
address 10.99.22.53
}
define host{
use docker-host
host_name pi504.in.thelinuxpro.net
alias pi504
address 10.99.22.54
}
### Service definitions for all docker hosts
define service{
use local-service,graphed-service
hostgroup_name docker-hosts
service_description PING
check_command check_ping!100.0,20%!500.0,60%
}
define service{
use local-service,graphed-service
hostgroup_name docker-hosts
service_description Docker Daemon
check_command check_nrpe!check_proc_docker
}
define service{
use local-service,graphed-service
hostgroup_name docker-hosts
service_description Disk: /
check_command check_nrpe!check_disk
}
define service{
use local-service,graphed-service
hostgroup_name docker-hosts
service_description Disk: /var/lib/docker/volumes
check_command check_nrpe!check_disk_docker
}
define service{
use local-service,graphed-service
hostgroup_name docker-hosts
service_description System: Load
check_command check_nrpe!check_load
}
define service{
use local-service,graphed-service
hostgroup_name docker-hosts
service_description System: Total Processes
check_command check_nrpe!check_procs
}
define service{
use local-service,graphed-service
hostgroup_name docker-hosts
service_description System: Zombie Processes
check_command check_nrpe!check_procs
}
define service{
use local-service,graphed-service
hostgroup_name docker-hosts
service_description System: Logged In users
check_command check_nrpe!check_users
}
define service{
use local-service,graphed-service
hostgroup_name docker-hosts
service_description DNS Query: A Record
check_command check_dns_a
}
### Service Definitions per docker host ### Service Definitions per docker host
define service{ define service{
use local-service,graphed-service use local-service,graphed-service

View File

@ -4,10 +4,3 @@ define hostgroup{
alias Docker Hosts alias Docker Hosts
members pi501.in.thelinuxpro.net,pi502.in.thelinuxpro.net,pi503.in.thelinuxpro.net,pi504.in.thelinuxpro.net members pi501.in.thelinuxpro.net,pi502.in.thelinuxpro.net,pi503.in.thelinuxpro.net,pi504.in.thelinuxpro.net
} }
### Hostgroup
define hostgroup{
hostgroup_name web-hosts
alias web Hosts
members kameronkenny.com,thelinux.pro
}

View File

@ -5,7 +5,7 @@
# NOTES: This config file provides you with some example object definition # NOTES: This config file provides you with some example object definition
# templates that are refered by other host, service, contact, etc. # templates that are refered by other host, service, contact, etc.
# definitions in other config files. # definitions in other config files.
# #
# You don't need to keep these definitions in a separate file from your # You don't need to keep these definitions in a separate file from your
# other object definitions. This has been done just to make things # other object definitions. This has been done just to make things
# easier to understand. # easier to understand.
@ -67,7 +67,7 @@ define host{
name linux-server ; The name of this host template name linux-server ; The name of this host template
use generic-host ; This template inherits other values from the generic-host template use generic-host ; This template inherits other values from the generic-host template
check_period 24x7 ; By default, Linux hosts are checked round the clock check_period 24x7 ; By default, Linux hosts are checked round the clock
check_interval 2 ; Actively check the host every 5 minutes check_interval 5 ; Actively check the host every 5 minutes
retry_interval 1 ; Schedule host check retries at 1 minute intervals retry_interval 1 ; Schedule host check retries at 1 minute intervals
max_check_attempts 10 ; Check each Linux host 10 times (max) max_check_attempts 10 ; Check each Linux host 10 times (max)
check_command check-host-alive ; Default command to check Linux hosts check_command check-host-alive ; Default command to check Linux hosts
@ -80,38 +80,44 @@ define host{
register 0 ; DONT REGISTER THIS DEFINITION - ITS NOT A REAL HOST, JUST A TEMPLATE! register 0 ; DONT REGISTER THIS DEFINITION - ITS NOT A REAL HOST, JUST A TEMPLATE!
} }
### Host Template
define host{
name docker-host # Windows host definition template - This is NOT a real host, just a template!
use generic-host
hostgroups docker-hosts
check_period 24x7
check_interval 3
retry_interval 1
max_check_attempts 3
check_command check-host-alive
notification_period 24x7
notification_interval 120
notification_options d,u,r
contact_groups admins
register 0
}
define host{ define host{
name web-host name windows-server ; The name of this host template
use generic-host use generic-host ; Inherit default values from the generic-host template
hostgroups web-hosts check_period 24x7 ; By default, Windows servers are monitored round the clock
check_period 24x7 check_interval 5 ; Actively check the server every 5 minutes
check_interval 2 retry_interval 1 ; Schedule host check retries at 1 minute intervals
retry_interval 1 max_check_attempts 10 ; Check each server 10 times (max)
max_check_attempts 3 check_command check-host-alive ; Default command to check if servers are "alive"
check_command check-host-alive notification_period 24x7 ; Send notification out at any time - day or night
notification_period 24x7 notification_interval 30 ; Resend notifications every 30 minutes
notification_interval 120 notification_options d,r ; Only send notifications for specific host states
notification_options d,u,r contact_groups admins ; Notifications get sent to the admins by default
contact_groups admins hostgroups windows-servers ; Host groups that Windows servers should be a member of
register 0 register 0 ; DONT REGISTER THIS - ITS JUST A TEMPLATE
} }
# We define a generic printer template that can be used for most printers we monitor
define host{
name generic-printer ; The name of this host template
use generic-host ; Inherit default values from the generic-host template
check_period 24x7 ; By default, printers are monitored round the clock
check_interval 5 ; Actively check the printer every 5 minutes
retry_interval 1 ; Schedule host check retries at 1 minute intervals
max_check_attempts 10 ; Check each printer 10 times (max)
check_command check-host-alive ; Default command to check if printers are "alive"
notification_period workhours ; Printers are only used during the workday
notification_interval 30 ; Resend notifications every 30 minutes
notification_options d,r ; Only send notifications for specific host states
contact_groups admins ; Notifications get sent to the admins by default
register 0 ; DONT REGISTER THIS - ITS JUST A TEMPLATE
}
# Define a template for switches that we can reuse # Define a template for switches that we can reuse
define host{ define host{
@ -158,8 +164,8 @@ define service{
is_volatile 0 ; The service is not volatile is_volatile 0 ; The service is not volatile
check_period 24x7 ; The service can be checked at any time of the day check_period 24x7 ; The service can be checked at any time of the day
max_check_attempts 3 ; Re-check the service up to 3 times in order to determine its final (hard) state max_check_attempts 3 ; Re-check the service up to 3 times in order to determine its final (hard) state
check_interval 2 ; Check the service every 10 minutes under normal conditions check_interval 10 ; Check the service every 10 minutes under normal conditions
retry_interval 1 ; Re-check the service every two minutes until a hard state can be determined retry_interval 2 ; Re-check the service every two minutes until a hard state can be determined
contact_groups admins ; Notifications get sent out to everyone in the 'admins' group contact_groups admins ; Notifications get sent out to everyone in the 'admins' group
notification_options w,u,c,r ; Send notifications about warning, unknown, critical, and recovery events notification_options w,u,c,r ; Send notifications about warning, unknown, critical, and recovery events
notification_interval 60 ; Re-notify about service problems every hour notification_interval 60 ; Re-notify about service problems every hour