Server Monitoring

There are multiple different systems that work together to support basic monitoring and metrics on across the servers. The server aspects all run on Alfred and take in data from clients on each of the systems.

The Core

The core system runs on using Sensu, an open source monitoring system. It is spread over multiple daemons and systems as defined below:

Dashboard

Port:8085
Password protected:
 Yes
Service:/etc/init.d/uchiwa
Configuration:/etc/sensu/uciwa.json

This display all the alerts and statuses of the servers. Anything that has been identified as a problem is shown here.

Graphite

Port:8086
API port:2003
Service:/etc/init.d/graphite-carbon.conf
Configuration:/opt/graphite/conf/carbon.conf

Handles the storage and display of server metrics. Graphs aren’t pretty or easy to deal with so an alternative graphing system is used.

RabbitMQ

Port:8087
Service:/etc/init.d/rabbitmq-server
Configuration:/etc/rabbitmq/rabbitmq.config

This is the queue manager that handles the output from the various client servers so needs to be internet accessible. All of the connections are protected through SSL.

Graphing

Although the Graphite service does provide graphing capabilities these are not easy to use or very nice to look at.

Grafana

The Grafana website contains an overview of all of the servers key metrics as well as more detailed dashboards covering these, along with others, in more detail. It is a javascript powered website so settings are kept in Nginx and in some configuration files in the directory.

Port:8088
Configuration:/srv/www/grafana/config.js
Nginx Configuration:
 /etc/nginx/sites-available/grafana.conf