Monitoring Architecture

Version 1.6.9 | Updated: August 08, 2026

Monitoring Architecture

Overview Diagram

Monitoring Architecture Overview

Data Flow

Data Flow Diagram

SNMP Trap Flow (Granular OID Mapping)

Each Prometheus alert rule has a unique oid label. When an alert fires, the flow is:

flowchart LR
    subgraph Prometheus["Prometheus"]
        RULE["Alert Rule<br/>oid: 1.3.6.1.4.1.65666.2.0.48"]
    end

    subgraph AM["Alertmanager"]
        ROUTE["Route to<br/>snmp-notifier webhook"]
    end

    subgraph SNMP["SNMP Notifier :9464"]
        PARSE["Extract oid label"]
        MIB["MIB Lookup<br/>FRIENDLY-TECH-ACS-MIB"]
        TRAP["Generate SNMP Trap"]
    end

    subgraph NMS["NMS / Receiver"]
        RX["Receive trap<br/>with unique OID"]
    end

    RULE -->|firing| ROUTE
    ROUTE -->|webhook + labels| PARSE
    PARSE --> MIB
    MIB --> TRAP
    TRAP -->|"SNMPv2c/v3<br/>UDP :162"| RX

Components:

  • Alert Rules (prometheus_rules.yml, hazelcast_rules.yml, acs_alerting_rules.yml) — 81 rules; 36 carry SNMP OID labels across 33 distinct trigger OIDs (+ paired oid_resolved; some OIDs shared across related alerts)

  • Alertmanager — routes firing alerts to snmp-notifier via webhook

  • SNMP Notifier (:9464) — extracts oid label from alert, generates SNMP trap with that OID

  • MIB file (alertmanager/mibs/FRIENDLY-TECH-ACS-MIB.txt) — maps OIDs to NOTIFICATION-TYPE definitions, mounted as external config

  • OID Mapping (alertmanager/mibs/OID-MAPPING.md) — human-readable OID reference

  • Thresholds UI (/alertmanager-config/thresholds) — edit alert thresholds without restart

OID Structure: 1.3.6.1.4.1.65666.2.0.<alert-id>

  • 1.3.6.1.4.1.65666 — Enterprise OID (Friendly Technologies, IANA PEN 65666)

  • .2.0 — ACS Notifications branch (ftAcsTraps)

  • .<alert-id> — unique per alert (1-99: infrastructure, 100-200: reserved for customer)

Server Types & Exporters

Server Type Exporters Ports

ACS/WildFly

node, process, cadvisor, jmx-hikari, jmx-jvm, Hazelcast native endpoint (embedded in ACS), ft-system, acs-metrics, subscription-api

9100, 9256, 9183, 5556, 5557, 9101, 8090, 8080

Hazelcast (standalone / docker)

node, process, cadvisor, jmx-hazelcast (JMX :9110 → serves :9101)

9100, 9256, 9183, 9101

MySQL

node, process, cadvisor, mysqld-exporter

9100, 9256, 9183, 9104

Oracle

node, process, cadvisor, oracledb-exporter

9100, 9256, 9183, 9161

ClickHouse

node, process, cadvisor, clickhouse (native endpoint)

9100, 9256, 9183, 9363

PostgreSQL

node, process, cadvisor, postgres-exporter

9100, 9256, 9183, 9187

API/HC1 (Angular UI)

node, process, cadvisor, provision, northbound, service, nginx-exporter, ft-device-network, ai-agent, ui-backend

9100, 9256, 9183, 8091, 9880, 8085, 9113, 8383, 8084, 8881

Windows

windows_exporter — registered with TYPE=NODE (node-style metrics on the same port); service=IIS drives the Windows/IIS dashboard

9100

Metrics Endpoints Reference

System Exporters (Docker containers)

servers.env TYPE Port Metrics Path Description

NODE

9100

/metrics

CPU, RAM, Disk, Network (node-exporter)

PROCESS

9256

/metrics

Host process stats (process-exporter)

CADVISOR

9183

/metrics

Docker container metrics (cAdvisor)

MYSQL

9104

/metrics

MySQL database metrics (mysqld-exporter)

ORACLE

9161

/metrics

Oracle database metrics (oracledb-exporter)

CLICKHOUSE

9363

/metrics

ClickHouse database metrics (native Prometheus endpoint)

POSTGRES

9187

/metrics

PostgreSQL database metrics (postgres-exporter)

NGINX_PORTALS

9113

/metrics

Nginx web server metrics (nginx-exporter)

WINDOWS

9100

/metrics

Windows system metrics (windows-exporter)

JMX Exporters (Docker containers)

servers.env TYPE Port Metrics Path Description

JMX_HIKARI

5556

/metrics

HikariCP connection pool metrics

JMX_JVM

5557

/metrics

JVM heap, GC, threads metrics

Native Application Endpoints

servers.env TYPE Port Metrics Path Description

JMX_HC

9101

/metrics

Hazelcast metrics: embedded-in-ACS via native :9101; standalone via the jmx-hazelcast exporter (JMX :9110 → :9101) — native alone omits isClusterSafe/MBean metrics

FT_SYSTEM

8090

/actuator/prometheus

FT System Spring Boot Actuator

ACS_METRICS

8080

/actuator/prometheus or /prometheus/metrics (auto-detected)

ACS application business metrics

SUBSCRIPTION_API

8080

/rest/subscription-metrics

Subscription API metrics (FTACS)

API Service Endpoints (Docker containers / Native Actuator)

servers.env TYPE Port Metrics Path Description

PROVISION_API

8091

/prov-portal/actuator/prometheus

Provision Portal API metrics

NORTHBOUND_API

9880

/iot-webservice/actuator/prometheus

Northbound API metrics

SERVICE_API

8085

/iot-webservice/actuator/prometheus

Service API metrics

FT UI Services Endpoints (Docker containers / Native metrics)

servers.env TYPE Port Metrics Path Description

FT_DEVICE_NETWORK

8383

/api/actuator/prometheus

FT Device Network Service (Spring Boot)

AI_AGENT

8084

/metrics

AI Agent (Python/FastAPI with prometheus_client)

UI_BACKEND

8881

/actuator/prometheus

UI Backend (when actuator enabled in Spring Security)

Monitoring Stack

Service Port Description

Prometheus

127.0.0.1:9090

Metrics storage & query

Alertmanager

127.0.0.1:9093

Alert routing & notifications (Email, Telegram, Slack, Teams, Webhook, SMS, SNMP)

SNMP Notifier

127.0.0.1:9464

Converts alerts to SNMP traps

Grafana

127.0.0.1:3100

Dashboards & visualization

Alertmanager Config UI

8085

Web editor for notification channels, thresholds and alert rules

Nginx

80/443

Reverse proxy, HTTPS, /grafana-ro/, /alertmanager/, /alertmanager-config/

Every service above except nginx and the Config UI binds the loopback only, and the stack runs in network_mode: host — so these are not ports you can reach from another machine, and not ports to open in a firewall. The browser talks to nginx on 443 and nothing else.

Grafana’s port is the one that catches people. It is 3100, hardcoded as GF_SERVER_HTTP_PORT in docker-compose.yml; there is no variable for it. GRAFANA_PORT in the stack .env is the external HTTPS port nginx listens on, default 443. Probing :3000 — Grafana’s upstream default, which this stack does not use — reports a healthy stack as down.

How the Config UI applies a change. Channel settings live in the stack .env, not in alertmanager.yml — Alertmanager renders that file from its template in the entrypoint, and SNMP Notifier turns its SNMP_* variables into CLI flags in the same way. Neither re-reads .env while running, so the UI restarts them over the mounted /var/run/docker.sock (SNMP Notifier only when an SNMP_* value actually changed). Both the socket and the files the UI edits are reached through the host docker group, whose gid bootstrap.sh records as DOCKER_GID. If a restart is not possible the save still lands in .env and the UI reports which service is still running the old settings. Full details live with the app itself, in prometheus-grafana-stack/alertmanager-config-ui/README.md.

Testing Services (Optional)

Service Port Profile Description

MailPit

8025

testing

Fake SMTP server with web UI for email testing

SNMP Trap Catcher

8089

snmp

SNMP trap receiver with web UI for testing

branch 1.6.9 · commit a862b45d92590285f229f80a6b5d2efabcb3b1b0 · page generated August 08, 2026