Monitoring via Nagios or Icinga

Self Hosted only

General

The pascom server includes a service called prometheus. It collects all the metrics of the system. These include, among other things, the utilisation of CPU, RAM and HDD. By means of a plugin, these values can be queried externally.

Grant external access to prometheus

Login to the server via SSH as admin and get root access. Stop the Controller Container. Edit the file /var/lib/lxc/controller/vars.json and set EXPOSE_PROMETHEUS to on. Start Controller Container again.

sudo su
cs container stop controller
vi /var/lib/lxc/controller/vars.json

"EXPOSE_PROMETHEUS": "on"

cs container start controller

Prometheus Nagios Plugin

To query the metrics a plugin is needed. This can be integrated via a check command.

https://github.com/prometheus/nagios_plugins

Examples

Check if CSD service is running

./check_prometheus_metric.sh -H 'https://_PASCOM-IP_:8443/prometheus' -q 'up{job="csd"}'  -n "Running CSD Service" -w 0 -c 0 -C "-k" -t vector -m eq

Check if database service is running

./check_prometheus_metric.sh -H 'https://_PASCOM-IP_:8443/prometheus' -q 'container_up{name="pg"}'  -n "Container PG up" -w 0 -c 0 -C "-k" -t vector -m eq

RAM usage (Critical < 90%, Warning < 75%)

./check_prometheus_metric.sh -H 'https://_PASCOM-IP_:8443/prometheus' -q '100-(node_memory_MemAvailable_bytes/node_memory_MemTotal_bytes*100)'  -n "Memory used in %" -w 75 -c 90 -C "-k" -t vector

Usage of the SYSTEM Disk (Critical < 3GB, Warning < 8GB)

./check_prometheus_metric.sh -H 'https://_PASCOM-IP_:8443/prometheus' -q 'node_filesystem_avail_bytes{mountpoint="/SYSTEM"}/1024/1024/1024'  -n "Number of free diskspace in GB" -w 8000 -c 3000 -C "-k" -t vector -m lt

Usage of the RAM Disk (Critical < 400MB, Warning < 200MB)

./check_prometheus_metric.sh -H 'https://_PASCOM-IP_:8443/prometheus' -q 'node_memory_MemAvailable_bytes/1024/1024'  -n "Number of free diskspace in MB" -w 8000 -c 3000 -C "-k" -t vector -m lt

CPU utilisation (load1) (Critical < 10, Warning < 5. Values are multiplied by 10)

./check_prometheus_metric.sh -H 'https://_PASCOM-IP_:8443/prometheus' -q 'node_load1*10'  -n "Load average 1m" -w 50 -c 100 -C "-k" -t vector

CPU utilisation (load5) (Critical < 6, Warning < 4. Values are multiplied by 10)

./check_prometheus_metric.sh -H 'https://_PASCOM-IP_:8443/prometheus' -q 'node_load5*10'  -n "Load average 5m" -w 40 -c 60 -C "-k" -t vector

CPU utilisation (load15) (Critical < 4, Warning < 3. Values are multiplied by 10)

./check_prometheus_metric.sh -H 'https://_PASCOM-IP_:8443/prometheus' -q 'node_load15*10'  -n "Load average 15m" -w 30 -c 40 -C "-k" -t vector

Updated on 11 Dec 2020