★ Enterprise Edition — not included in the Community Edition. Compare editions

Prometheus Setup¶

The shipped prometheus.yml, how it's mounted, retention, and the known target corrections.

Prometheus (prom/prometheus:v2.51.0) collects metrics from every worker service and the infrastructure exporters. The active configuration is monitoring/prometheus/prometheus.yml, mounted read-only into the container together with the rules directory:

YAML

# docker-compose.yml (shipped)
prometheus:
  image: prom/prometheus:v2.51.0
  volumes:
    - ./monitoring/prometheus/prometheus.yml:/etc/prometheus/prometheus.yml:ro
    - ./monitoring/prometheus/rules:/etc/prometheus/rules:ro
    - prometheus_data:/prometheus
  command:
    - '--config.file=/etc/prometheus/prometheus.yml'
    - '--storage.tsdb.path=/prometheus'
    - '--storage.tsdb.retention.time=30d'
    - '--web.console.libraries=/etc/prometheus/console_libraries'
    - '--web.console.templates=/etc/prometheus/consoles'
    - '--web.enable-lifecycle'
  ports:
    - "9090:9090"        # production binds 127.0.0.1:9090

prometheus.yml — As Shipped¶

YAML

global:
  scrape_interval: 15s
  evaluation_interval: 15s
  scrape_timeout: 10s

rule_files:
  - "/etc/prometheus/rules/*.yml"

alerting:
  alertmanagers:
    - static_configs:
        - targets: ['alertmanager:9093']

scrape_configs:
  - job_name: 'prometheus'
    static_configs: [{ targets: ['localhost:9090'] }]

  # Mail services
  # (postfix-exporter/dovecot-exporter jobs are commented out in the real file
  # since 2026-08-30 — neither exporter exists in any compose file; the rspamd
  # job followed on 2026-08-31 — this build's controller serves no /metrics)

  # Worker services — targets use internal container ports
  - job_name: 'api'
    metrics_path: /metrics
    static_configs: [{ targets: ['api:8080'] }]
  - job_name: 'webhooks'
    static_configs: [{ targets: ['webhooks:8081'] }]
  - job_name: 'rate_limiter'
    static_configs: [{ targets: ['rate_limiter:8082'] }]
  - job_name: 'monitoring'
    static_configs: [{ targets: ['monitoring:8085'] }]
  - job_name: 'tracking'
    static_configs: [{ targets: ['tracking:8086'] }]
  - job_name: 'analytics'
    static_configs: [{ targets: ['analytics:8085'] }]          # container port (host-mapped to 8087)
  - job_name: 'dashboard'
    static_configs: [{ targets: ['dashboard:8088'] }]
  - job_name: 'archiver'
    static_configs: [{ targets: ['archiver:8083'] }]
  - job_name: 'encryption'
    static_configs: [{ targets: ['encryption:8084'] }]
  - job_name: 'queue_manager'
    static_configs: [{ targets: ['queue_manager:8090'] }]
  - job_name: 'rag'
    static_configs: [{ targets: ['rag:8090'] }]                # container port (host-mapped to 8091)
  - job_name: 'storage_usage'
    static_configs: [{ targets: ['storage_usage:8092'] }]
  - job_name: 'delivery_optimizer'
    static_configs: [{ targets: ['delivery_optimizer:8088'] }]
  - job_name: 'jmap'
    static_configs: [{ targets: ['jmap:8098'] }]
  - job_name: 'migration'
    static_configs: [{ targets: ['migration:8099'] }]
  - job_name: 'autoconfig'
    static_configs: [{ targets: ['autoconfig:8100'] }]
  - job_name: 'caldav'
    static_configs: [{ targets: ['caldav:8101'] }]
  - job_name: 'templates'
    static_configs: [{ targets: ['templates:8089'] }]
  - job_name: 'url_protection'
    static_configs: [{ targets: ['url_protection:8090'] }]
  - job_name: 'oauth'
    static_configs: [{ targets: ['oauth:8091'] }]

  # Infrastructure exporters
  - job_name: 'mysql'
    static_configs: [{ targets: ['mysql-exporter:9104'] }]
  - job_name: 'redis'
    static_configs: [{ targets: ['redis-exporter:9121'] }]

(A kafka-exporter:9308 job is present but commented out — no kafka-exporter service exists yet.)

[!WARNING] Targets must use container ports, not host-published ports — the service-to-service hop inside the compose network never crosses the host port mapping. Two jobs originally got this wrong (analytics targeted 8087 instead of 8085, rag targeted 8091 instead of 8090) and scraped nothing; both were fixed, the dead postfix/dovecot exporter jobs commented out, and the missing templates/url_protection/oauth jobs added on 2026-08-30.

Retention¶

Retention is fixed by the compose command flag: --storage.tsdb.retention.time=30d (there is no env var for it). To change it, edit the command: list; you can also add --storage.tsdb.retention.size=10GB — whichever limit is hit first triggers cleanup.

Rough sizing for this stack's metric volume: ~2 GB for 30 days at the shipped 15 s interval. For longer horizons use remote write (Thanos/Mimir) rather than growing local retention.

Exporters¶

Only two ship:

YAML

mysql-exporter:
  image: prom/mysqld-exporter:v0.15.1
  environment:
    - MYSQLD_EXPORTER_PASSWORD=${DB_PASSWORD}
  command:
    - '--mysqld.address=mysql:3306'
    - '--mysqld.username=${DB_USER:-mailuser}'
    - '--no-collect.slave_status'   # no replica; the app user lacks REPLICATION CLIENT

redis-exporter:
  image: oliver006/redis_exporter:v1.58.0
  environment:
    - REDIS_ADDR=redis://redis:6379

If you add postfix/dovecot/node exporters, give them the service names the commented-out scrape jobs in prometheus.yml already expect (postfix-exporter, dovecot-exporter), uncomment those jobs, and add a node job — then restore the matching commented-out rules in rules/mail_alerts.yml.

Reloading Configuration¶

--web.enable-lifecycle is set, so a hot reload works after config edits:

Bash

curl -X POST http://localhost:9090/-/reload    # dev; use 127.0.0.1 on the prod host
# or
docker compose restart prometheus

Validate before reloading:

Bash

docker exec prometheus promtool check config /etc/prometheus/prometheus.yml
docker exec prometheus promtool check rules /etc/prometheus/rules/mail_alerts.yml /etc/prometheus/rules/backup_alerts.yml

Verifying Targets¶

Open the Prometheus UI — http://localhost:9090 in development, ssh -L 9090:127.0.0.1:9090 in production (the port is loopback-bound there).
Status → Targets — every worker job should be UP.
Expect postfix and dovecot DOWN (no exporters), and analytics/rag DOWN until the port fix above is applied.

If a worker target is DOWN: is the container running (docker compose ps)? Does docker exec prometheus wget -qO- http://<service>:<port>/metrics work from inside the network? Remember metric names come back prefixed with the service name — see Prometheus Metrics.