Architecture Overview

Energy Monitor runs as a set of components built around the Naemon monitoring core. Event broker modules, a message queue and several data stores connect them. This chapter describes what runs on a standard installation, what each component does and how data moves between them.

Read it before you size a deployment or debug a check that the core executes but the interface does not show.

Components

Component	Location	Role
Naemon core	`/opt/naemon`	Schedules and executes checks, evaluates states, sends notifications
Naemon Livestatus	`/usr/lib64/naemon/naemon-livestatus/livestatus.so`	Event broker module exposing the live state of the core over a socket
LMD	service `lmd`	Aggregates one or more Livestatus sources into a single endpoint for the interface
Merlin	service `merlind`	Distributes checks between nodes and replicates state in clustered installations
Statusengine broker	`/opt/statusengine/lib/libstatusengine.so`	Event broker module publishing every core event to Gearman
Gearman	service `gearmand`, bound to `127.0.0.1`	Message queue between the broker module and the worker
Statusengine worker	`/opt/statusengine/worker`	Consumes the queue and writes events and metrics to MySQL and OpenSearch
OpenSearch	port `9200`	Stores historical events and performance metrics
MariaDB	service `mariadb`	Stores configuration, interface state and Merlin data
Redis	service `redis`	Cache
Ninja	`/opt/energy-monitor/ninja`	Main web interface, PHP on the Kohana framework
Nacoma	`/opt/energy-monitor/nacoma`	Configuration interface, writes the Naemon object configuration
Monitor API	`/opt/monitor_api`	REST API, Laravel, documented with Swagger
PNP4Nagios	`/opt/energy-monitor/pnp`	Performance graphs based on RRD, fed through `npcd` and `rrdcached`
NagVis	`/opt/energy-monitor/nagvis`	Status maps
BPI	`/opt/energy-monitor/business`	Business process modelling and SLA thresholds
CMDB	`/opt/energy-monitor/cmdb`	Asset inventory and risk assignment, introduced in 1.6
Intelligence	installed by the `intelligence` package	Backend for AI Correlations
Geomap, Topology	separate packages	Geographic and topological views
Notify	package `notify`	Notification integrations, among others Microsoft Teams, Discord, Jira, Slack, Telegram
Agents	`/opt/agents`	Build environment for the Windows and Linux monitoring agents
Apache with php-fpm	services `httpd`, `php-fpm`	Serves every web component

Data flow

Check execution

Naemon schedules and executes checks according to the object configuration in /opt/naemon/etc. The core processes each result and decides on state changes and notifications.

From the core to the interface

The Livestatus broker module exposes the current state over a socket. LMD reads that socket and presents a single endpoint to Ninja, which queries it for host and service lists, dashboards and Monitor Wall.

In a distributed installation LMD aggregates several Livestatus sources, so the interface shows a combined view without querying each node separately.

Ninja translates its own query language into Livestatus queries. The lsfilter module parses the query into an abstract syntax tree and compiles it to one of several outputs: Livestatus, SQL, PgSQL, statistics or native PHP. The same mechanism serves list views, Monitor Wall and dashboard widgets.

From the core to history and metrics

The Statusengine broker module is loaded into Naemon through /opt/naemon/etc/naemon/module-conf.d/statusengine.cfg and publishes every core event to Gearman on 127.0.0.1. The Statusengine worker consumes that queue and writes to the statusengine database in MariaDB and to OpenSearch.

The interface reads historical events, the event log and performance metrics from OpenSearch. The core holds only the current state.

Performance graphs

Performance data also goes to RRD files through npcd and rrdcached, and PNP4Nagios renders them. The Livestatus module is configured with pnp_path=/opt/energy-monitor/pnp/perfdata, which lets the interface link a service directly to its graph.

Configuration

Nacoma keeps the configuration in MariaDB and generates Naemon object files from it. The interface collects changes and applies them when you save the configuration. This is why it shows a pending change counter and a changelog.

Monitor API writes to the same configuration store. It covers configuration CRUD, /config/change, BPI, commands and the filter endpoints. Reports, graphs, the event log, dashboards, maps, the scheduler, agents, integrations, AI, CMDB and permissions have no API equivalent and are available only through the interface.

Commands from the interface

The interface writes acknowledgements, downtimes and forced checks to the Naemon external command pipe. The core picks them up on its next command cycle, so they take effect with a short delay.

Databases

Database	Used by	Contents
`monitor`	Ninja	Interface state: dashboards, saved filters, reports, user settings
`nacoma`	Nacoma	Configuration objects and the change log
`merlin`	Merlin	Node and distribution data
`statusengine`	Statusengine worker	Event and check history

The statusengine database and its user are created during installation by repo/files/statusengine.sql.

Web layer

Apache serves every component from a single virtual host under separate aliases:

Alias /monitor              /opt/energy-monitor/ninja
Alias /ninja                /opt/energy-monitor/ninja
Alias /business             /opt/energy-monitor/business
Alias /cmdb                 /opt/energy-monitor/cmdb
Alias /new_api              /opt/energy-monitor/new_api
Alias /monitor/energy/nacoma /opt/energy-monitor/nacoma
Alias /monitor/energy/pnp   /opt/energy-monitor/pnp
Alias /monitor/images       /opt/monitor/share/images
Alias /reportsImg           /opt/energy-monitor/ninja/reportsImg

The main interface embeds several components in an iframe instead of rendering them itself: Nacoma configuration forms, CMDB and the agent builder. Such a component can return an error inside the frame while the surrounding page still returns HTTP 200.

Services

An installation enables the following units:

naemon          monitoring core
merlind         distribution and replication
lmd             Livestatus aggregation
gearmand        message queue, bound to 127.0.0.1
statusengine    event and metric worker
opensearch      event and metric storage
mariadb         databases
redis           cache
httpd, php-fpm  web layer
npcd, rrdcached RRD performance data
bpi-status      BPI status service
nachos          agent and scan services
sssd            directory integration
postfix         mail delivery for notifications

The installer also enables monitor-network-autoscan.timer, which drives scheduled network discovery.

Version notes

The Statusengine worker runs on PHP 7.2 from the Remi repository, installed under /opt/remi/php72, with the gearman, igbinary and redis extensions. It talks to OpenSearch through the Elasticsearch PHP client in the 7.11 line. The web interface runs on a different PHP version.