ActiveMQ Artemis

ActiveMQ Artemis is Apache’s high-performance message broker — the successor to ActiveMQ Classic, common wherever JMS-style messaging runs. Operationally it worries you in the same places as any broker: queues backing up, consumers going missing, and — Artemis-specifically — an address hitting its memory limit and switching into its configured full-address policy (paging, blocking or dropping). This system type covers all three.

Key: artemis · applied automatically to services emitting artemis-prefixed metrics.

Getting the telemetry

Artemis exposes metrics through the community artemis-prometheus-metrics-plugin, which registers a Prometheus endpoint on the broker’s web port. Scrape it with the Collector’s prometheus receiver:

otel-collector-config.yaml (receivers)

receivers:
  prometheus/artemis:
    config:
      scrape_configs:
        - job_name: artemis
          scrape_interval: 30s
          metrics_path: /metrics
          static_configs:
            - targets: ["artemis:8161"]

The full walkthrough — enabling the plugin in broker.xml, the export pipeline, reading the queue and dead-letter signals — is in Monitor ActiveMQ Artemis queue depth.

What Sluicio detects

Metrics whose names start with artemis — artemis_message_count, artemis_consumer_count, artemis_address_memory_usage, and the rest of the plugin’s export.

Starter checks

Check	Condition	Severity	Why it matters
Queue backlog	More than 5,000 messages on a queue (artemis_message_count)	Warning	Consumers aren’t keeping up — the earliest sign an integration is falling behind.
No consumers	A queue has messages but zero consumers (artemis_consumer_count)	Warning	Nothing is draining the queue — usually a crashed or disconnected consumer.
Address memory high	An address’s memory usage approaching its configured limit	Warning	At the limit, Artemis applies the address-full policy — paging to disk, blocking producers, or dropping messages.

Best-effort metric names

These checks are written against the standard names exported by the Artemis Prometheus metrics plugin. Metric names can shift between plugin and broker versions — if a check never fires, compare its metric name against your broker’s actual /metrics output and adjust the rule on the service.

Tuning notes

• Backlog threshold. As with any queue, derive the number from the drain rate of the specific queue, and give the rule a duration so a normal burst doesn’t page anyone.
• Address memory is the check most worth an early look: what “high” means depends entirely on your max-size-bytes settings per address in broker.xml. Align the threshold with the addresses that block producers rather than page.
• Dead-letter growth isn’t a starter check here because the DLQ is just another queue in Artemis — add your own rule on artemis_message_count for your dead-letter address (default DLQ), alerting on any sustained increase. The Artemis guide shows the pattern.